Sector report - pleinetete

May 6, 2008 - Open source develops on a bottom up approach, i.e. it first conquers the ... compete with SAP Business ByDesign and Business Objects on-demand. ... I 6. 1. The open source software business model. The open source model has a ...... --Java (TM) 2 Platform Standard Edition Development Kit 6.0 32-bit.
1MB taille 38 téléchargements 623 vues
Sector report 6 May 2008

Europe

Software - IT services

137 117

Outperformance

97 77 avr-05

août-06

janv-08

DJ Stoxx Technology Rel.DJ Stoxx

Open source SaaS: the next killer application?

Source: Natixis Securities

n Open source shapes up as the new eldorado in the software world as is shown in

transaction prices, the latest example being the acquisition of MySQL by Sun, in January 2008, for $1bn, i.e. just under 12x 07 sales. n Growth rates in the segment are well over those seen in the proprietary software

segments. For the applicative layer, open source was still only worth $2.9bn in 2006 (i.e. 3% of the total market) but this should rise 43% per year between 2006 and 2011 to reach $17.2bn, i.e. 14% of the market in this timeframe. n The

success of open source software lies mainly with interoperability, independence, code flexibility and a TCO (total cost of ownership) way under that for proprietary solutions. The TCO gap between Linux and Windows is 28% over three years, it is 10x less than for a proprietary solution in databases.

n We do not see major risk on the large accounts installed base for proprietary ERP

and Business intelligence vendors over the next 3 years. However, with the growing maturity of open source solutions, coupled with the adoption of the SaaS model, we believe that the competition will be stiff at the bottom of the midmarket and that it could affect SAP. n The large generalist IT services providers are still barely present on the open

source niche and are allowing open source integrators to develop. Players like Capgemini and Atos Origin strike us as insufficiently attentive to this source of growth future. They could react once this model matures by acquiring players at a high price.

Natixis Securities Bloomberg access

www.sec.natixis.com NXSE

Company

Rating

Atos Origin Capgemini Dassault Systèmes SAP Steria LogicaCMG Sopra Group GFI Informatique Devoteam Sector average

Add Add Add Add Add Reduce Add Reduce Add

Price

Target

€ 38.77 € 39.33 € 39.70 € 32.52 € 22.60 p 120.00 € 49.67 € 4.58 € 21.55

€ 44.00 € 45.00 € 47.30 € 39.50 € 25.00 p 113.00 € 59.50 € 4.50 € 25.00

P/E (x) 2008 12.7 12.6 18.8 17.1 7.8 11.4 9.4 8.2 10.5 12.0

EV/EBIT (x) 2009 9.5 11.3 16.9 14.6 6.5 10.0 8.3 7.5 8.5 10.3

2008 9.3 6.3 11.2 12.6 6.1 9.5 6.2 6.2 4.7 8.0

2009 7.0 5.4 10.0 10.5 5.0 8.3 5.2 5.5 3.3 6.7

EV/Turnover (x) 2008 0.5 0.5 3.1 3.6 0.5 0.7 0.6 0.4 0.4 1.1

Distribution of this report in the United States. See important disclosures at the end of this report.

2009 0.5 0.5 2.7 3.1 0.4 0.7 0.5 0.4 0.3 1.0

Sector report

Table of Contents Investment summary

4

1. The open source software business model

6

The various types of economic model

6

Major advantage: low cost of possession

9

Effective business model

11

Open source: SWOT analysis

18

2. The open source market

20

Strong worldwide growth and bottom-up invasion

20

Taking over from major accounts... beyond 2010

24

France, a market at the leading edge

25

3. Leaders in emergent open source

27

The infrastructure layer well penetrated

27

The applications layer: an offering gaining in maturity

38

4. Impact of open source on traditional players

48

Repercussions for proprietary vendors?

48

Repercussions for IT services providers?

49

5. Appendices

52

The open source model

52

Different types of open source licences

52

The intrinsic qualities of open source software

56

Software - IT services

I3

Sector report

Investment summary Open source looks like a new eldorado given the growth rates of players and current transaction multiples. The success of open source software lies mainly with interoperability, independence, code flexibility and a TCO (total cost of ownership) way under that for proprietary solutions owing to the absence of licence cost. According to IDC, the world open source market chalked up sales of $18bn in 2006, and should reach $35bn in 2008, i.e. a CAGR of 25%. For Gartner, sales stemming from open source reached $18.4bn in 2006 and are likely to grow 30.4% to $24.1bn in 2007 and 30.6% to $31.4bn in 2008. The market share for open source in the software segment amounted to 10% in 2006 and will probably climb to 12% in 2007 then 14% in 2008 (source: Gartner). Open source develops on a bottom up approach, i.e. it first conquers the infrastructure layer via Linux for servers, and is expanding increasingly in applicative layers (office, management). For the applicative layers, open source was only worth $2.9bn in 2006, i.e. 3% of the total market, and should rise with a 2006/2011 CAGR of 43% to generate $17.2bn in 2011. IDC assesses ERP open source market share at 7% in 2008 for a total market estimated at €26.7bn. ERP open source vendors are likely to penetrate the SME market initially as the licence cost forms a high proportion of the overall budget for the solution. Hence the ERP open source threat for large accounts is weak at three years but needs taking seriously on the TPE segment. At SAP, it may enter into direct competition with: 1/ Business One and even Business ByDesign, the future SaaS version solution which, according to the group, has to provide extra sales of $1bn in 2010 (i.e. 8% of 2010e sales at SAP), a target which has in fact just been postponed by 12-18 months, 2/ the business intelligence offer for the midmarket. Open source BI is still far from being a standard in IT systems despite the clear advantage in cost terms (about 30% of the total cost savings for a decisional project). As for the ERP segment, the main brake on expansion for open source BI in large accounts is the cost of migration for existing proprietary solutions. In BI as in ERP, when the licence price is a determining criterion and the potential client still lacks solutions, open source then becomes competitive. For this reason, the midmarket will be the choice segment for the development of the open source BI and ERP. We think that the open source/SaaS duo could become a killer application in the future in the midmarket. As with Sun’s takeover of MySQL, there could be convergence between the proprietary vendor world and that of open source pure players. It will be gradual by market segment as a function of the adoption maturity. It will be mainly initiated by proprietary vendors eager to maintain their hegemony while meeting their clients requirements. Open source being emergent in the world of ERP and Business Intelligence, SAP is likely to be gradually interested. It could make its first open source acquisition in the domain of SaaS for the midmarket, the segment most likely to adopt these technologies and which will thus directly compete with SAP Business ByDesign and Business Objects on-demand. The large proprietary vendors, which possess both the installed base and the financial means, cannot stand aside from a market growing fast, likely to compete with their core business and

Software - IT services

I4

Sector report

dominated by small pure players. Companies, all unlisted, like Jaspersoft, SugarCRM, ERP5, Compiere, could in due course become choice targets for SAP. Table 1: High transaction multiples Date January 2008 August 2007 Sept. 2007 Feb. 2006 June 2006 Oct. 2005 May 2005 Nov. 2003 August 2003

Proprietary Sun Citrix Yahoo Oracle Red Hat Oracle IBM Novell Novell

Open source MySQL Xen Source Zimbra Sleepycat JBoss InnoDB Gluecode Suse Ximian

Segment Databases Virtualisation Messaging Databases Application server Storage engine Application server Operating system Graphic envir.

Transaction price $1bn for sales 07 of 85me (+55%) $500m for sales of $50me in 2008 $350m for sales 07 of $25-35m $35m for sales 07 of $7m $350m for sales 05 of $23m Nd $100m for sales 05 < $1m $210m for sales of $40me $50m for sales of $1m

Multiples 11.8x rev. 10x rev. 11.7x rev. 5x rev. 15x rev. nd ns 5x rev. ns

Sources: Companies, Natixis Securities

The open source software market offers an interesting alternative and a growth relay for IT services providers. However, the it is still too limited for it to be a major source of sales for services providers. It is dominated by the pure player integrators and specialists in open source support services on the market. However, with 2% of 2007 sales at Capgemini and under 0.5% of Atos Origin, its presence within large services providers now looks much too weak. In due course, the competitive situation should shift leading to the acquisition of open source integrators by the large IT services providers. IT services providers will not acquire open source vendors as this would amount to developing solutions which compete with their large proprietary vendor partners.

Software - IT services

I5

Sector report

1. The open source software business model The open source model has a number of strengths it can boast of: scalability/flexibility, productivity, reliability, compatibility with standards, function guarantee, independence, duration of developments and architectures. Its main strength is still its low total cost of ownership (TCO). Necessary costs (technical skills, implementation of solutions) do go some way towards counterbalancing the absence of licence costs.

The various types of economic model From a technical perspective, software comprises a source code and an executable code. The move from one version to another is achieved largely thanks to a ‘compilation’, which translates the source code into a code executable by the computer. The term open source software does not mean that it is free but available source code. An open source software package is software which is delivered with its source code while proprietary software is delivered in executable form. The expression ‘open source’ refers to the freedom for users to execute, copy, distribute, study, modify and improve the software. More precisely, it refers to four types of liberty for the software user: 1/ the freedom to execute the programme for all usages; 2/ the freedom to examine the functioning of the programme, and to adapt it according to needs; 3/ the freedom to redistribute copies; 4/ the freedom to improve the programme and to publish improvements, so that the entire community benefits. What changes with open source is the positioning of the value which is located in the service and no longer in the code. Open source vendors and integrators are remunerated on the basis of maintenance, integration, skills provision and the tailoring of the code for clients. Chart 1: The open source business model The indirect services or valuation model

The value added distribution model

Business Model The double licence or non-commercial source licence

The ‘pooling’ model

Source: Fabernovel

Software - IT services

I6

Sector report

Integration services The clients buy a service provision from one or several open source software packages, which consists in: − Installation/deployment of the software. − The parametered implementation of the software. − Training in the utilisation of the software. − The achievement of specific developments: creation of integrated software packages based on a open source development platform (LAMP: Linux, Apache, PHP, MySQL). The vendor commercialises the usage of the software as a function of the benefit it provides for the end user. Asssurance services for maintenance On top of the integration services, open source vendors offer maintenance and support services which ensure problem correction for the client according to a reaction time stipulated in the support contract. This is the case for example for the OSSA (Open Source Software Assurance) programme put in place by Linagora as part of the project to revamp the IT set-up for the French tax authorities. Generally, there are three levels of support: − Level 1: corrective, adaptive and scalable maintenance. − Level 2: permits a guaranteed response time in maintenance. − Level 3: offers guaranteed priority response times, plus 24h/24 access. Services for the pooling of the economic cost Two cases of pooling of developments on a pre-paid basis: − The case of the service provider which has control over the product coherence and puts several clients in contact to offer to finance a complete software package or the addition of new functions. The pooling of prepaid developments is then done as part of user clubs. Very large companies and large accounts appreciate sharing their modes of operation and their demands with players on the same infrastructure. Example of users of this model: ERP5 and Nuxeo (ECM) offer the pooling of developments and maintenance. − The case of ecosystems around the software: vendors together build a product and service offer. The more players there are in the grouping, the more the solutions provided by the grouping become credible in the eyes of deciders. Non vendor intermediaries (IT players only selling services) can easily become a player in the development of the solution, thereby ensuring that the ecosystem lasts. Example of users of this model: OpenOffice.org (which comprises Sun, IBM, Novell, Google, Intel, etc.), the Eclipse foundation (project supported by IBM, Borland, Sybase, etc.), the Apache foundation (IBM, HP, etc.).

Software - IT services

I7

Sector report

Turnkey solutions The clients pay to obtain a turnkey solution, for example a solution (firewall + antivirus + proxy + anti spam) or extended pre-parametered software, for example a OpenOffice.org deployment kit and self-training modules. The service provider plays the integrator role: it undertakes all the work in operation, equipment deployment and develops add-ons (such as the management graphic interface for example) and offers a ready to use solution. The double licence The vendor offers a downloadable community open source version, and a corporate version for maintenance service subscriptions. This is the case for MySQL Community Server, the community version (free downloading and available source code) and the MySQL Enterprise version, connected to maintenance services (source code available, but the product is linked to maintenance subscriptions, no executable binary available in free downloading). Table 2: The double licence models Company

Segment

MySQL Pentaho Jaspersoft OpenXchange AliaSource Xensource Zend Zimbra SugarCRM OpenBravo Alfresco Compiere Red Hat Jboss Talend

SGBD BI BI Groupware Groupware Virtualisation IDE Groupware CRM ERP ECM ERP OS Server app ETL

Licence model Community & commercial version ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?

Sale model Subscriptions ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?

Add-ons sales ? ? ?

?

? ?

Sources: Conecta, Natixis Securities

In other cases, the vendor offers a community version with an available source code and a pay version without a source code. This is the example of Zend which supplies a platform to execute PHP (Zend framework) as a base for its open source offer, with subscriptions for maintenance (corporate version), and offers third party development tools under proprietary licence (graphic interface, debugging, code optimiser) to develop around the software. In general, these third party tools are not strictly necessary for the use of the software base but they improve the productivity of developers or make the complexity of the software easier to access for less technical profiles. On top of subscriptions to updating services, vendors can offer premium services such as the sale of modules (add-ons). The vendor offers paying extensions distributed under proprietary licences. The base of the code is still the open source licence but the proprietary extensions (source code not available) are paid for and meet the challenges with ramp up faced by large structures.

Software - IT services

I8

Sector report

Major advantage: low cost of possession In the appendix we have detailed the intrinsic qualities of open source software: scalability/flexibility, productivity, reliability, compatibility with standards, guaranteed functioning, independence, duration of developments and architectures. But the most emblematic of theses is the low total cost of ownership (TCO). By definition of the open source model, the TCO for a solution based on open source software is lower than for a proprietary solution. A survey, undertaken in 2004 by Cybersource, thus compared the TCO for a Linux solution (Red Hat Enterprise Linux) and Windows. The study was based on an IT system consisting of 250 users over a period of 3 years (with maintenance). Table 3: Assessment of the TCO for a RHEL platform (Red Hat) compared with a platform Windows (Microsoft) 250 users with 3 years of maintenance MS Windows platform Licence + support 3 years Quantity Price USD Antivirus 250 9,475 Windows server 2003 9 35,991 MS Information Server 2 0 MS Commerce Server 1 19,999 MS ISA Standard Server (pare-feu) 1 1,499 MS SQL Server 1 19,999 MS Exchange Server 1 3,999 MS XP pro 250 74,750 MS Visual Studio .NET 3 7,497 MS Office 2003 3 99,750 Adobe Acrobat 250 598 Adobe Photoshop 2 1,298 Client Acces Licences Windows Server 2003 2 1,675 Client Acces Licences MS Exchange 2003 250 16,750 MS software Assurance Program for servers 74,934 MS software Assurance Program for workstations 136,498 Total 504,712

RHEL platform Support 3 years Quantity RHEL AS 1 RHEL ES 3 RHEL ES 5 RHEL WS 5 RH Desktop Satellite Pack 1 RH Desktop Extension Pack 4 Apache (Web server) incl. in RHEL AS Squid (proxy server) incl. in RHEL AS MySQL incl. in RHEL AS Iptable (firewall) incl. in RHEL AS Postfix (mail server) incl. in RHEL AS Kdevelop (IDE) incl. in RHEL AS GIMP incl. in RHEL AS OpenOffice incl. in RHEL AS OS commerce 1 Total

Prix USD 7,191 5,235 4,485 40,500 42,000

99,411

Employees over 3 years Administrators Consultant Training Total Total ($) Gains with open source (%) Gains ($)

3 1 1

444,000 45,000 30,000 519,000 1,023,712

Administrators Consultants Training Total

3 1 1

474,000 135,000 30,000 639,000 738,411 28 285,301

Source: Cybersource (2005)

The resultant of the Cybersource report shows a TCO lower for Linux by 28% over 3 years on an unchanged hardware configuration. Another survey conducted by MySQL highlights that in database management systems (SGBD), the analysis of the TCO shows an average cost over 3 years 10x higher than for proprietary solutions compared with open source.

Software - IT services

I9

Sector report

Table 4: Calculation for the TCO between Oracle, MySQL, Microsoft, Sybase, IBM $ SGBD Version Code Model of price License $ (unit) Support (unit)

Microsoft

Sybase

IBM

Oracle

MySQL Enterprise MS SQL Server 2005 Gold Enterprise Open source Proprietary Server CPU 0 24,999 2,995 5,000

MySQL

Sybase ASE 15 Enterprise Proprietary CPU 24,995 4,999

IBM DB2 v9 Enterprise Proprietary CPU 36,400 7,280

Oracle 10g Enterprise Proprietary CPU 40,000 8,000

Costs Licence 12 CPU Support 3 years

0 53,910

299,988 180,000

299,940 179,964

436,800 262,080

480,000 288,000

TCO TCO (3 years)

53,910

479,988

479,904

698,880

768,000

426,078 89% 8x

425,994 89% 8x

,644,970 92% 12 x

714,090 93% 13 x

TCO gains Gains MySQL x Source: MySQL

In the evaluation of a SGBD, the difference in costs must however be put in the balance with other critical factors: − The interoperability and integration capacities, i.e. at the administration level, development and necessary human skills. − The oval value for the company. The TCO just measures the cost and not the value. To measure the generated value we also need to factor in variables such as productivity, quality, supportability, maintainability, ease of utilisation, flexibility and duration. In addition, the exit ticket in the case of large accounts is very high and it is difficult for an IT director to play the competition. What is true for new projects is no longer true when there is a long track record with a proprietary vendor. The Cybersource report also indicates the administrative costs for the Linux platform which are higher than for Windows (the administration model for the open source platform is not unified, which necessitates the development of numerous scripts and additional resources). The consulting and training costs are also higher. But beyond the simple price calculation, other factors needs including. In particular, the Cybersource model parts from the unlikely assumption that there is no track record in the company: no servers, no peripherals, no applications on servers or work stations, and no skills. Including these items, 28% of the benefits of the open source solution will be largely overrun by the training cost for administrators and users, by the migration of applications, and by the renewal of peripherals not supported by Linux. One of the challenges for IT directors is the predictability of costs: proprietary vendors do not guarantee tariff models followed over time (billing according to the number of users who have access to the base, according to the number of CPUs, the number of servers). Moreover, vendors proprietary tend to accelerate cycle obsolescence for products and launch expensive upgrades on the market which are not mature (800 corrections for Oracle Applications 10g).

Software - IT services

I 10

Sector report

Table 5: TCO for open source software versus proprietary software Category Installation Integration in the IS

+/+

Equipment Licences Security

+ + +

Training Assistance

=

Evolution

+

Archiving

+

Comments The installation tools are more complex than for proprietary software The respect of open standards guaranteed by the open source software permits to ensure a lower cost of insertion in the IS Open source software requires cheaper equipment and network configurations By definition, the licence cost is nil The open source development mode provides better assurance of the security of developments, evolution (updates) and data uptake IT employees trained in proprietary systems need to be re-trained, lifting the transition cost Assistance in terms of open source software is being built up, which makes it at least as expensive as its proprietary counterpart The respecting of standards permits the separation between software bricks, which can be developed independently The respecting of open standards ensures that archiving formats are respected

Source: Rapport Carcenac (2001)

Effective business model The takeoff for free software has allowed the emergence of new economic models and the creation of new companies like Red Hat, MySQL and has taken part in the mutation of older companies like Novell. These modes have also led to the takeoff of French open source software companies such as Talend and Linagora, companies we met with on several occasions while writing this report. New ecosystems The free software market above all consists of service oriented open source vendors and more recently services providers specialised in open source software or ‘SSLL’ (Sociétés de Service en Logiciels Libres ). This market is still highly freemen and theses specialists generally have to partner IT services providers of stuffiness size to meet ambitious calls for tender. Chart 2: Interactions between players IT services companies

Open source service companies

Open source market

Standard market

Vendors

Source: PAC (2007)

The open source vendor model is based on a strong community of developers. The relation with the community over the long term becomes vital for code evolution. The services to clients are undertaken directly or via a network of integrators. The two main differences with

Software - IT services

I 11

Sector report

the proprietary model (vendor/integrators/client) lie with importance of the community and the lesser dependence vis-à-vis IT services providers. Chart 3: Open source ecosystem

Client

Integrator & vendor

Developers community

Source: Natixis Securities

This market is structured around a services offer, that is to say the certification and maintenance of blocks of open source software. As the code is freely available it is at the level of service quality that value added is created. Is the model as effective as at proprietary vendors? While the return on investment for a vendor of proprietary software is based on the sale of licences, the free distribution of open source software does not allow this procedure to be used (the placing under GPL licence does away with pure licence revenues). Hence the question arises: will the revenues generated by the services offer suffice to make up for the lack of licence revenues? In the open source model, the sale of licences is replaced by the sale of maintenance around software bricks, under the form of yearly subscriptions on usage (1-3 years and according to the number of users). For example, Red Hat generated sales of $533m in 2007, of which 86% from subscriptions to updates for the enterprises using Red Hat Enterprise Linux and 14% from services (training, consulting). The revenues from the subscriptions to maintenance services replace the licences in the proprietary model. With open source, the client has the source code and the licence revenues are replaced by more flexible and reactive services: installation services, personalisation, configuration, maintenance. Levers for the open source model Viral diffusion is one of the strengths of open source software. The principle is simple: as the GPL licence allows free copying of the code, the sources are made available on internet, notably via P2P networks.

Software - IT services

I 12

Sector report

This allows prompt diffusion worldwide and the vendor can benefit from: − A large installed base. − A large number of contributors (information on bugs, contributions to developments, improvement suggestions, translations). − A large number of ‘prescribers’ which eases the penetration of IT systems in companies, by commencing installation on non strategic servers by IT staff users. With the proprietary model, the vendor spends a large proportion of turnover on promotional and marketing costs (20% of sales at SAP, i.e.; 30% of costs, 22% of sales at Oracle). With open source, clients come directly to the vendor, which internationalises faster by leaning on the communities (development driven by demand and not supply). The commercial strategy in open source software is based on its distribution canal which is the downloading of the source code. The aim is then to obtain a maximum of downloads and to monetise the installed base (subscriptions to the maintenance services for the bricks). Compared with proprietary vendors, open source vendors can thus eliminate sales and marketing costs to begin with. Access to the market is thus eased and the marginal cost of acquisition for a client in open source is lower than with the propritary model. Chart 4: Pre-sale cycle for open source versus proprietary Open Source

Proprietary

Offer

t Lis

Demos

End contract

rs ide rov ep vic Ser

Offer

rs ide rov ep vic Ser

End contract

Pilot training Short List Self

Downloading Forums Support

e vic ser

Specifications Sales documentation (brochure, CD)

Pre-sales payment on licences

Pre-sales payment on services

Source: Altic.org

Open source offers a pilot project procedure (for a cost for the client at some 10% of the project budget) which permits modelling of the critical processes (assessment of the gap, test of the application potential, first level of learning). This approach offers the advantage of validating the project scope, testing, identifying and reducing risks. Speed of availability on the market (ramp up) Only then come the levers on the pooling of the development of the code which enables open source vendors to refine functionalities as effective as those for proprietary solutions and more rapidly.

Software - IT services

I 13

Sector report

In cost terms, R&D spending for a proprietary vendor and a open source vendor do not show a strong trend. The R&D effort between open source vendors varies according to the business model. However, the supply on the technologies market is more effective at open source vendors. With the proprietary model, the amortisation of R&D costs passes via the exclusivity of lines of code. In open source, the cost of the development of a software package is shared with, aside from the internal R&D teams: − The R&D common to the ‘free’ world, which comprises work by thousands of developers in the open source community (who work for example, on the Linux core). They supply some software bricks that the vendor will then integrate. The cost of this R&D is then nil for the vendor. − The specific external R&D which comprises contributor developers not paid by the vendor but which contribute to the improvement of its distribution. The number of these contributors varies a great deal depending on the reputation of the vendor and its open source ethic. Chart 5: The open source community around the Mandriva project

Open source projects

Community

Mandriva -Developers -Sales force -Marketing -Management

Professional users

Prescrbers / Contributors

Individual users

Mandriva contributors

Open source partners

Source: Mandriva

− Other open source projects: the reutilisation of components from other open source software helps secure and speed up the development of the project (for example Red Hat Enterprise Linux v 5.0 integrates the Xen project for the virtualisation). − The large software vendors and constructors, which finance the technological portage of open source solutions towards their own architectures. For example, IBM has financed the portage of Red Hat Enterprise Linux v5.0 on its mainframe servers by obtaining the certification at level 4+ of Evaluation Assurance Level. This is an international standard for the security of IT systems for the various US governmental administrations. IBM also takes part in the development of the OpenOffice suites and highlights tools like PostgreSQL, Drupal, Talend, Xen, Pentaho and Compiere. − IT services providers, which offer tailored services around the application (for example, Alcôve employs 6 developers specialised on the distribution of Linux Debian which hands over their contributions to the community). The French company Linagora intends to put in

Software - IT services

I 14

Sector report

place a network of integrator partners which will decline by industry its flagship solution OBM in messaging and open source collaborative work. − A community of clients, who pools their resources to finance specific developments. This is the case for OpenTrust, which sells ‘mutualised’ open source software such as IDX-PKI (security) for Inserm, the CEA and GAN Patrimoine. The clients are united in a club of contributors and agree to pay their developments into the community. The greater the adoption, the greater the number of contributors. The greater the number of users, the longer will be the software’s lifespan. As a result, the evolving and corrective maintenance cost diminishes as it is shared by a larger number of clients. Chart 6: Model of ‘pooling’ by clients

Source: Fabernovel

This pooling of R&D costs permits speed of product provision (ramp-up) on the market which is faster than with the proprietary model, and real time maintenance. The difference is clear as some proprietary vendors no longer innovate as rapidly as the open source model like OS Vista. Whereas it will have taken 5 years for Microsoft to release Vista, GNU/Linux Ubuntu offers a new version of its distribution every six months.

Software - IT services

I 15

Sector report

Chart 7: Breakdown of costs at a proprietary vendor and an open source BI vendor Business Objects

¨Pentaho Cost of sales 25%

Sales & marketing

Cost of sales 26%

24%

Sales & marketing 47%

G&A R&D

10%

18% G&A

R&D

10%

40%

Sources: Business Objects, Pentaho, Natixis Securities

Quality of source codes Historically, open source is born of the user’s frustration at not being able to correct problems in the software they have bought as the code source is not available. With open source, any developer can correct a programme. The platform for the management of users’ returns put in place by the vendor permits: − Developers to report problems encountered and to supply a corrected version of the programme. − Users not only to report problems encountered but also the functionalities they seek to incorporate. The project developer then takes on board the information to improve the following version of their product. These tools then permit a continual improvement of the quality of the code as well as ensuring that the code does not shift contrary to users’ desires.

Software - IT services

I 16

Sector report

In conclusion, the economic model for open source thus looks virtuous, notably thanks to its viral nature, flexibility, reactivity, and code quality. However, though the model allows barriers to entry to be lowered and to spread faster than the proprietary model, can the open source vendor retain its advantage in that the competition can gain access to the source code and develop a rival project as Oracle did with Oracle Enterprise Linux? It should be pointed out that generally an open source vendor offers a version for the community and another version for the company. The code for the community part is in GPL, diffusible and reproducible. But the most refined version of code of the company version is not in GPL. The code is available but non reproducible, which allows the vendor to protect itself (modified GPL licence). Chart 8: Proprietary model (left) versus open source (right)

22% 50%

14% 38%

50%

17%

8%

50%

7%

2%

5%

Sales &

G&A

18% Licenses Services

Cost of

R&D

Sales &

sales

Mktg. Sales

Costs

G&A

Earnings

Services

Cost of

R&D

sales

18% Earnings

Mktg. Sales

Costs

Sources: OpenBravo, Natixis Securities

Software - IT services

I 17

Sector report

Open source: SWOT analysis The two tables below summarise the strengths and weaknesses of the open source model compared with the proprietary model. They highlight the strengths of open source which are the code flexibility, the TCO, the time to market, interoperability, a growing number of applications and opportunities such as the backing of equipment manufacturers and services providers, the public sector, development in the midmarket, ramp up for the SaaS model. Amongst the weaknesses and threats which hang over the model, note the weight of the legacy, admin costs and the lack of maturity in some segments. Table 6: SWOT analysis for open source Strengths

Weaknesses

Code flexibility, viral buzz, scalability, TCO (cheaper hardware, no licence costs), verticalised functionalities, time to market, security, production code quality, interoperability (open standards, multi platform/architecture medium), independence, long code duration, pooled procedures, much weaker marginal acquisition cost for the client. Mature open source solutions: open source middleware (Linux, Apache, Squid, JBoss, Tomcat, GlassFish, MySQL, PostgreSQL, PHP, Perl, Python, Eclipse, Joomla, Drupal, SPIP).

The great number of open source licences (GPL, LGPL, BSD, EPL, MPL; Apache Licence Software, etc.) constitutes a risk of non interoperability: danger of persistence and permissiveness.

Mature open source applications for SMEs: BI (Jaspersoft, Talend, Pentaho, SpagoBI); ERP (Compiere, OpenBravo, Tiny ERP, ERP5); CRM (SugarCRM); ECM (Alfresco, Nuxeo); Groupware/portal/Web2.0 (OpenXchange, eXoplatform, Liferay, Xwiki, OBM); Identity management (FederID, OpenTrust); Administration (Nagios). Software producers focus on services, not products.

The open source project lacks maturity for critical applications, lacking vertical modules, support/competence, and offering limited documentation (e.g. Open Office, certain open source ERPs). : Administration of open source solutions is more complex.

The line is blurred between community version and corporate version open source (on a pay basis). Open source projects can be copied (e.g. Oracle’s Unbreakable Linux, copy of RHEL). Free open source and the idea that whatever is given away necessarily performs less well than a paid-for solution. Linux experts cost more and are harder to find. Migration costs towards open source tools. Optimised hardware for proprietary code, lack of certification.

Sources: Natixis Securities, Aful, Forrester

Software - IT services

I 18

Sector report

Table 7: Analyse SWOT of open source Opportunities

Threats

Growing competence of European engineers, greatly involved in the development of Pressure from proprietary vendors to promote a single star solution by software open source projects. segment. Substantial presence of open source tools in the development of Web 2.0. The large integrators offer a hub of competence (Capgemini, Atos Origin, Bull, Steria, Sopra, Unilog, Osiatis, etc.). IT services providers to large European accounts are less sensitive to open source layers, whereas some large US accounts have adopted open source thanks to support from such players as Sun, HP and IBM. Support from hardware vendors (IBM, HP, Dell, Asus, Acer). Subsidies from Brussels for European companies’ R&D is putting a brake on the adoption of open source by SMEs. European CIOs tend to mistrust proprietary licences and prefer the personalised services of software integration. Proprietary vendors are cutting their licence prices, licence costs on large projects have become modest versus total cost. Public sector support: Europe’s strong commitment to setting up open standards (e.g. the SIP standard in the telecoms sector). The large media players seem hostile to the emergence of a non commercial shared knowledge system. This could slow the development of open source technologies in Business to Consumer. 80% of commercial software will contain an open source component in 2010 (Gartner) Software patents. In Europe, software cannot be patented because of the methods and algorithms it uses (the same is true in mathematics). This has prompted the large software producers to push the principle of patenting software by claiming to foster innovation. Defenders of open source prove the contrary: since the IT innovation cycle is very short, patented software puts a freeze on innovation by prohibiting access to a method, instead of stimulating it. Concentration of software makers, the search for independent producers, a best-ofbreed approach. The Olivennes remit and the DADVSI law. This French copyright law harms the principle of technology’s neutrality (P2P), i.e. the principle that a given technology is in itself neither good nor bad, but the use made of it is either righteous or blameworthy. This law brings to bear legal insecurity on the creators of open source software. Java has moved to GPL, and future developments will be based on open source tools. Midmarket (less than 100 machines) and emerging countries (Brazil, Africa, China, India). Parallel development of the SaaS model. Source: Natixis Securities

Software - IT services

I 19

Sector report

2. The open source market The open source market has already reached a certain degree of maturity, more particularly in the fields of infrastructure (with Linux, MySQL, JBoss, Apache, Eclipse). It is emerging in the field of business transactional applications such as ECM, ERP, CRM, and also in Business Intelligence, with companies such as Jaspersoft and Talend. The last few years have also seen the growth of services based around open source, services offered both by open source software vendors and free software services companies. Chart 9: Adoption of open source by software layer

Directory Services

Enterprise Applications

Development Tools

Maturing

Most Mature

Content Management

Integration Services Process Management

Server OperatingSystem Security

Virtualization

Application Servers

Collaboration Search RDBMS

Enterprise Service Bus (ESB)

Embryonic

Emerging

Growing

Source: Gartner (2007)

Strong worldwide growth and bottom-up invasion The world open source market is growing from the bottom up, in other words it first conquered the IT system infrastructure layer through Linux for servers, and is increasingly spreading to the tool and application layers. The world open source market represented sales of around $18bn in 2006 (sources: IDC and Gartner). It is likely to continue to grow strongly in the next few years (+25% per year between now and 2008 for IDC and around 30% per year for Gartner). The market share of open source in the software segment was 10% in 2006 and should grow to 12% in 2007 then 14% in 2008. According to Gartner, the open source infrastructure layer accounted for 15% of the total market in 2006 with sales of $15.5bn and should grow by 28% in 2007. By way of comparison, sales generated by proprietary software in the infrastructure layer amounted to $86.7bn in 2006, and are expected to advance by 10% in 2007. In 2011, Gartner expects

Software - IT services

I 20

Sector report

open source infrastructure software to generate sales of $46bn, i.e. a 2006/011 CAGR of 24.3%, with a market share of 27%. Today, there is not a single area in the infrastructure market that does not have a high quality open source solution: databases (MySQL, PostgreSQL, Ingres), application servers (JBoss, JOnAs), server supervision tools (Nagios) and web servers (Apache). Chart 10: Impact of open source software on the infrastructure software layer worldwide ($m) 170 000

115 000

60 000

5 000 2004

2005

2006 Open Source

2007

2008 Proprietary

2009

2010

2011

Total

Source: Gartner (February 2007)

Geographically, the USA and western Europe, which are home to the majority of open source developer communities, are driving growth of the open source infrastructure market, with market shares of 40% and 30% respectively. In 2011, the USA is expected to generate open source sales of $18.2bn (2006/2011 CAGR of 25.3%) and western Europe sales of $12.2bn (2006/2011 CAGR of 24.4%). In the applications layer, open source only generated sales of $2.9bn (i.e. 3% of the total market) in 2006, and is expected to generate a 2006/2011 CAGR of 43%, bringing sales to $17.2bn in 2011, or 14% of the total market.

Software - IT services

I 21

Sector report

Chart 11: Impact of open source software on the application layer worldwide ($m) 125 000 100 000 75 000 50 000 25 000 0 2004

2005

2006

2007

Open Source

2008 Proprietary

2009

2010

2011

Total

Source: Gartner (February 2007)

The impact of open source decreases as one moves further up the value chain to the application layers. Not only have developer communities not achieved sufficient critical size to produce a reliable code, but the solutions involve senior executives of companies who are less receptive to technological aspects. Infrastructure layers were easier to break into as open source developers were mainly speaking to IT people, who are generally open to innovative solutions. Table 8: Maturity of open source technologies Adoption of open source layers in SI 2,500 employees SMEs 2,500 > employees > 500 VSBs < 500 employees Total

2007e 30 30 15

2011e 40 51 30

CAGR (%) 7.5 14.2 19.0

75

121

12.7

Sources: IDC, Gartner, SAP

The recent launch of SAP’s on-demand solution (SAP Business ByDesign) illustrates this desire to conquer the IT systems of small businesses.

Taking over from major accounts... beyond 2010 To gauge the impact of open source on major accounts, the investments already made need to be taken into account. In the last 20 years, major accounts have made fundamental investments: 1/ networks and office communication tools (messaging, extranet, intranet); 2/ ERP (finance, logistics, human resources, production); and 3/ customer relations and, more recently, business intelligence. Thus, open source will not be able to make headway at major accounts unless it fits in with legacy solutions resulting from past investments. Within this framework, proprietary software bricks and open source software bricks will have to fit together. Open source standards could facilitate this cohabitation. For major accounts, the decision to opt for an open source or proprietary technology has to take into account the viability of the supplier, especially when the technology is strategic for the organisation. What counts is therefore operating efficiency, even though the panorama of costs and expertise also has importance. Operating efficiency also means greater transparency, and many chief information officers (CIOs) refuse to be technologically dependent (black box software), and reject the monopoly situation of proprietary software providers. For CIOs, open source has already reached an initial stage of maturity, i.e. the community of developers in certain projects is sufficiently large and well organised to actively support development of the code and ensure user support. For example, we may cite EADS subsidiaries, which are using ERP5 and which fit in perfectly with a proprietary architecture. Conversely, in CAD and PLM, open source software is not yet sufficiently mature, as the community is not sufficiently dense. According to the IT club for large French corporations CIGREF (Club Informatique des Grandes Entreprises Françaises), open source will naturally take over from proprietary software, thanks to the quality of the code, which results from an ecosystem in which 200,000 of the world’s best code writers constantly improve the quality of open source software. Last, interoperability and costs are also decisive factors in the choice of software. The CIGREF also

Software - IT services

I 24

Sector report

emphasises that open source meets a demand for transparency from IT departments, which want to reduce the risk of being reliant on a supplier (major accounts are no longer willing to undergo the migrations of versions imposed to correct bugs), and want to have a choice of solutions and responsibility for the software bricks.

France, a market at the leading edge France has become one of the most dynamic countries for open source software in the world, and could return to the software innovation race while IT departments are becoming increasingly important in the creation of value by companies. The potential is due to: 1/ strong involvement in the main communities; 2/ increasingly broad expertise and offerings; and 3/ projects with worldwide implications. According to PAC, the open source software and services market in France represented €440m (i.e. 1.5% of the total French software market) in 2007, up 70% compared with 2006. PAC forecasts growth of 50% on this market in the next few years. Chart 14: The open source software market in France 3 500 3 000 2 500 2 000 1 500 1 000 500 0 2005

2006

2007 2008 2009 Administration Other sectors

2010

2011

Source: PAC (2007)

The first clients for open source are either major accounts in the public sector, often driven by budget cuts, or scientific and university institutions, which are historical users of open source software. The promises of massive cost reductions, renewed technological independence and development of local services have resulted in strong political support for developing use of open source software. The French administration devoted 9% of its IT budget to open source developments in 2006 (compared with 6% in 2005), i.e. €900m. Markess International forecasts growth of more than 12% per year in 2008/09. For example, all the machines at the French National Assembly are equipped with open source software. They integrate the Linux/Ubuntu OS, the Firefox internet browser and the OpenOffice office suite. Moreover, the ministry of culture, the ministry for equipment and the agriculture ministry, as well as the gendarmerie have also chosen to switch to OpenOffice. Previously, government departments used proprietary software bricks, which prevented them from correcting the security shortcomings in sensitive applications, for example. The choice of open source for government administrations is therefore not just a question of cost, as the Software - IT services

I 25

Sector report

share of licences remains marginal in the administration’s overall IT budget. Rather a strong concern with independence is what led government administrations to prefer solutions that do not make the stability of their organisation subject to Oracle corrective patches. The project that provided the impetus for open source software in the French administration was the Copernic project in the tax administration (Direction Générale des Impôts, DGI), which began in 2005. The €1bn project involving 800 in-house staff and more than 1,000 outside services providers, is aimed at federating the IT systems of the DGI and the Treasury (Trésor Public) and creating e-services for users. The technological policy of the project focused on interoperable and standard (SOA) software bricks, so as to favour independence, sustainability and the ability to evolve over the long term. The DGI therefore chose Linux as the operating system for its servers, JBoss (the J2EE open source application server acquired by Red Hat in 2006) for its application servers and Nagios to supervise all of its architecture. The DGI called on the grouping of Capgemini-Bull-Linagora for the support, maintenance and services relating to open source bricks. Capgemini was charged more precisely with management of the change and Bull with server infrastructure. Linagora supplied its third party maintenance methodology applied to free software. The task of third party maintenance applied to free software is to correct anomalies, adapt the solution and also handle the evolution of the solution (make available new functions). Atos Origin managed the JBoss project. The heads of IT working on the Copernic project noted that managing a portfolio of Oracle solutions cost three times as much as the portfolio currently comprising 150 free software solutions.

Software - IT services

I 26

Sector report

3. Leaders in emergent open source Open source solutions have begun to penetrate infrastructure layers and are gradually progressing towards users, via the applicative layers. Several pure players, more or less mature, comprise the open source community. At this stage, there are still no global players in open source, each being specialised by market category. Red Hat has shown the example on the operating systems segment. In enterprise applications, notably in the SME segment, Jaspersoft and Pentaho, in BI, SugarCRM, Compiere and OpenBravo, in the ERP are highly promising players. Inside three years, they should become serious rivals to the large propriety vendors in the midmarket segment. Table 10: Sales for representative open source players worldwide Company Novell Red Hat MySQL Ingres Sourceforge Zend SugerCRM HP1 Open Trust Linagora2 Mandriva Open Wide Nuxeo Smile NetAktiv Proformatique 1

Type Vendor Vendor Vendor Vendor Mixed Vendor Vendor Constructor Vendor Mixed Vendor Integrator Vendor Integrator Integrator Vendor

05 sales $1,039m $196m $30. 6m $10m $25m €3.1m €5.4m $4m €1.9m €352K €100K

06 sales $967m $278m $50me $25m3 $43. 6m $15m $7.5m3 nd €10m €4m (€8.5m 07) €5.7m €4.5m €3.2 / 3.5m €9.1m €500K €300K

In 2006, HP estimates it made sales of €25m on the distribution of Linux Debian. This is the sole distribution to have generated substantial sales, but HP also sells Novell SUSE and Red Hat Linux servers.

2

Linagora merged in 2007 with Alia.

3

$50m in 2007 for Ingres, $15m for SugarCRM, $523m for Red Hat

Sources: Companies

The infrastructure layer well penetrated The infrastructure layer is the intermediary layer of communication between several applications distributed in an IT network. We are interested in the open source infrastructure layer: 1/ operating systems; 2/ middleware (Web servers, application servers) ; 3/ databases; 4/ open source development platforms (Java, PHP). Infrastructures software turns on servers and is thus dependent on the type of equipment architecture developed by constructors and the growth for the equipment rate in servers. The servers market (equipment and operating system) made sales in 2006 of $52.6bn. In operating systems, Linux comprises some 15% (versus 10% in 2005) of market share after

Software - IT services

I 27

Sector report

Unix and Windows. The breakthrough for GNU/Linux as an operating system is partly due to the interoperable nature of architecture processors/OS couples, while other proprietary systems are not compatible with as many architectures. Apache leadership in web servers A Web server is a server which sends requests respecting the client server communication protocol HyperText Transfer Protocol (HTTP). According to Netcraft, Apache is the open source Web server which turns the largest number of sites in the world since April 1996. In June 2007, Netcraft collated 122,000,635 internet sites, 58% of which operate with an Apache server. Chart 15: Market share in servers August 1995 – June 2007 80%

Apache Microsoft Sun Lighttpd Google NCSA Other

40%

Mar2007 Jun2007

Sep2005 Mar2006 Sep2006

Sep2004 Mar2005

Mar2004

Sep2003

Sep2002 Mar2003

Sep2001 Mar2002

Mar2001

Mar2000 Sep2000

Mar1999 Sep1999

Mar1998 Sep1998

Mar1997 Sep1997

Sep1996

Mar1996

Sep1995

0%

Source: Netcraft

The Apache Software Foundation (ASF) is a decentralised community of developers which works on open source projects. Apache projects are characterised by a collaborative development mode based on the consensus and by a non-copyright GPL Apache open source licence. This Apache licence authorises modification and distribution of the code in any form (open source and commercial, IBM http Server for example is based on the Apache web server). Apache mainly operates on the Unix type operating systems (GNU/Linux, BSD and UNIX) and Windows. It is thanks to its uncontested leader position within the community of developers, that Apache can consistently improve its open source code software. Apache’s major presence on the Web servers market is due to the success of the development platform LAMP (Linux, Apache, MySQL, PHP) for the web site. LAMP designates all the open source software enabling internet site servers to be built. Even though the authors of each of the programmes did not work together to build the LAMP platform, this combination of software has spread (Youtube operates using LAMP) owing to the low overall costs and the presence of all these components in Linux distributions. Vendors can alter and add functions to the source code, then re-condition and even sell this code with their own applications.

Software - IT services

I 28

Sector report

Chart 16: The LAMP architecture

Source: Natixis Securities

The LAMP solution has made an impact as it permits great stability for the HTTP server. Netcraft data allow the comparison of the stability of sites hosted under a Linux platform with a proprietary platform. The period of time since the system reboot on the free site (linux server) averages 133 days and that for Business Objects.com (Windows 2000 server) 25 days, clearly illustrating the stability of the Linux solution. Open source operating systems dominated by Red Hat According to Gartner, the world market in operating systems for servers could reach $12.3bn in 2008, i.e. an increase of 4%. The OS servers market is dominated by Windows (40.5% market share in 2007). Linux for server is credited with 13% market share in 2007. Linux will see the steepest growth rate out to 2011 with CAGR estimated at 8%. At that stage, Linux will have market share estimated at 16%.

Software - IT services

I 29

Sector report

Table 11: Operating profit systems for servers market Windows (Server) Linux (Server) Sun Solaris IBM System z IBM AIX HP-UX Other proprietary OS IBM System i Other proprietary Unix Total

2006

2007

2008

2009

2010

2011

CAGR 2006/2011 (%)

4,219.3 1,353.5 1,351.2 960.8 916.4 849.9 705.1 327.6 250.5

4,788.3 1,569.8 1,442.5 973.1 942.4 898.2 625.9 328.5 243.1

5,132.0 1,787.4 1,462.8 955.9 951.5 888.7 569.2 321.4 230.7

5,415.3 1,927.5 1,464.9 942.0 946.4 878.7 523.4 315.9 216.7

5,677.2 2,049.3 1,460.0 924.5 932.3 864.6 486.5 310.2 201.6

5,882.7 2,146.9 1,451.6 910.2 930.3 847.1 457.0 305.6 185.3

6.9 9.7 1.4 -1.1 0.3 -0.1 -8.3 -1.4 -5.9

10,934.3

11,811.8

12,299.6

12,630.9

12,906.1

13,116.8

3.7

Source: Gartner (Octobre 2007)

Open source OS have benefited from support from the community, which has provided some sector-specific adaptations (telecoms, public sector, finance, etc.) to codes. In addition, OS have been able to adapt to changes both in server hardware (blades, compatibility with IA-64 and RISC architectures) and software, including virtualisation (RHEL includes open source Xen technology). GNU/Linux has been buoyed in particular by support from large players like IBM, which has little by little sidelined AIX and Z/OS and featured GNU/Linux on the System z (RHEL 5 and SLES 5 have both been endorsed by IBM). The market for open source databases alone was worth $15.2bn in 2006, with Linux accounting for 15.5% of this total. Table 12: Operating systems strictly for database servers 2004

2005

2006

04 mkt. share (%)

05 mkt. share (%)

06 market share (%)

Chg. 05 (%)

Chg. 06 (%)

5,476.6 3,934.6 782.8 1,255.8 647.9 393.6

5,337.3 4,291.7 1,408.3 1,312.3 606.4 367.6

5,293.6 5,241.6 2,352.9 1,414.7 577.7 333.2

43.8 31.5 6.3 10.1 5.2 3.2

40.1 32.2 10.6 9.8 4.6 2.8

34.8 34.5 15.5 9.3 3.8 2.2

-2.5 9.1 79.9 4.5 -6.4 -6.6

-0.8 22.1 67.1 7.8 -4.7 -9.3

12,491.3

13,323.5

15,213.7

100.0

100.0

100.0

6.7

14.2

$m Unix Windows Server Linux zOS/OS390 Others iOS/OS400 Total

Source: Gartner (Juin 2007)

Growth in the number of databases running on Linux (proprietary or open source) has been driven by such large vendors as Oracle (Real Application Clusters 10/11g available on RHEL) and Teradata (SLES). Gartner estimates that Linux OS penetration of database management system (DBMS) servers will reach 50% five years out. What is pushing CIOs to move to Linux is the low TCO, combined with platform flexibility and hardware compatibility. The market for open source servers is dominated by Red Hat (Red Hat Enterprise Linux 5.0) and Novell (Suse Linux Enterprise Server 10). Red Hat supplies open-source bricks (OS and middleware) as well as assistance, training and consulting services. Its best known product is Red Hat Enterprise Linux (RHEL), an OS for enterprise servers.

Software - IT services

I 30

Sector report

Red Hat delivers RHEL on an annual subscription basis with one year of maintenance. It is licensed under the amended GPL; source packages are accessible and can be compiled. There are charges for support and maintenance with all versions of RHEL. Red Hat supports each version of the software for seven years and offers upgrades every 18-24 months, plus daily updates via the Red Hat Network. Table 13: The Red Hat offer covers all material architectures Onboard systems

RTOS (IBM)

Office stations

RHEL =

Windows

AS

Company servers

OSX (Mac)

ES Control stations

WS

High performance calculation

Desktop

UNIX

Telecoms infrastructure Source: Red Hat

The latest version, RHEL 5, is configured as a server and workstation. Table 14: The Red Hat offer Red Hat Enterprise Linux5.0 Enterprise Linux AS (Advanced Server) Enterprise Linux ES (Edge Server) Enterprise Linux WS (Workstation) Red Hat Desktop Multi/OS

Segments Core company servers and critical IT systems. Mid-sized network servers Technical and/or large scale work stations Simple work stations

Source: Red Hat

Support services vary depending on the required responsiveness to incidents. For instance, the AS Premium service is recommended when service interruption cannot be tolerated; in the telecoms sector, Nokia runs its network data centres on RHEL. Table 15: The Red Hat support offer REHL AS Premium AS standard ES Premium ES Standard

Price OS/year $ 2 499 1 499 799 349

Characteristics Support 24x7, incident response in under 1 hour, for any type of problem Support 5x12, incident response in under four hours, for any type of problem Support 5x12, intervention in under 4 hours, for any type of problem Support 30 days, intervention in under 2 days, for any type of problem

Source: Red Hat

Compared with proprietary solutions using the same technology, RHEL 5 offers a platform that is endorsed by the major software vendors and manufacturers along with a TCO including capital costs (system design, purchase of hard- and software, software migration, prototyping/tests, training and updates) and maintenance costs that is, according to an IDC report, 49% below the TCO for Unix over three years.

Software - IT services

I 31

Sector report

Table 16: TCO REHL vs UNIX over 3 years Cost per user ($) Equipment Software Employees Others TCO

Linux (RHEL) 1 077 1 058 942 4 423

Unix 3 060 2 627 4 507 4 874

7 500

15 068

Source: IDC (2002)

Red Hat also uses Red Hat Exchange (RHX), an online exchange platform, to distribute its solutions. Its goal is to create an open source software ecosystem around RHEL 5 (more than 3,000 applications certified in 2007). The platform could be a good way for Red Hat to assess demand for open source technologies and plan potential acquisitions. The RHX platform is built on REHL5 - JBoss - MySQL and features open source bricks from third parties like JasperSoft (open source leader for BI), Alfresco (open source leader in ECM), Compiere (ERP), Zimbra and Openfire (messaging and collaboration), Zmanda (backup), CentricCRM and SugarCRM, GroundWork and Zenoss (system supervision). Open source database management systems (DBMS) The database management system (DBMS) market generated total sales of $17.6bn in 2006, for a 12.2% increase according to Gartner. It is expected to expand by 8.4% a year between now and 2011, buoyed in large part by the open source segment. Forrester estimates that the open source database market was worth $400m in 2006 (technical support and maintenance) and predicts that this will increase to $1bn in 2008 and then to $4bn in 2010. The market is dominated by Oracle (the “incumbent”, for which DBMS accounted for 45% of 2006 sales via the flagship Database 9g-10g-11g solutions), IBM (DB2) and Microsoft (Microsoft SQL Server). IBM is still in the number two slot but, with growth trailing that of the market (8.3%), Microsoft is gaining ground and could pull in front within the next two years.

Software - IT services

I 32

Sector report

Table 17: SGBD market $m Oracle IBM Microsoft Other vendors Teradata Sybase Progress Software CA Fujitsu InterSystems Hitachi Software AG Pervasive Software Open source vendors Total

2005

2006 05 market share 06 market share

Chg. 05 (%)

Chg. 06 (%)

6 238.2 3 927.6 2 562.0 687.9 467.6 449.9 265.5 319.3 219.1 164.6 177.8 142.6 30.6 102.5

7 168.0 4 254.0 3 245.0 649.7 494.2 486.7 292.3 213.8 208.0 179.9 171.3 154.2 26.9 139.8

39.6 24.9 16.3 4.4 3.0 2.9 1.7 2.0 1.4 1.0 1.1 0.9 0.2 0.7

40.5 24.1 18.4 3.7 2.8 2.8 1.7 1.2 1.2 1.0 1.0 0.9 0.2 0.8

5.3 2.4 14.2 -4.7 5.7 17.5 4.2 0.9 -2.7 7.2 7.1 -0.4 -13.7 47.0

14.9 8.3 26.7 -5.6 5.7 8.2 10.1 -33.0 -5.1 9.3 -3.7 8.1 -12.1 36.3

15 755.3

17 683.7

100.0

100.0

5.6

12.2

Source: Gartner (Juin 2007)

The open source DBMS market can be divided into three categories: − ERP and CRM applications (85% of open source solutions, examples include MySQL, Firebird and Apache Derby). − OEM solutions, widely used in the telecom sector (e.g. MySQL and Oracle). − DBMS for data warehouses (e.g. MySQL and Ingres). The key players are: − MySQL (double GPLv2.0 and proprietary licence; there is a charge for the licence if MySQL is embedded in a proprietary product, otherwise it is under the GPL). − Ingres (I ntelligent Graphic Relational System, formerly owned by Computer Associates, GPL). − Apache Derby (Apache 2.0 licence), also known as Cloudscape, an IBM fork for recovering maintenance on the installed base. − Firebird (Mozilla licence), the only independent DBMS, no partnership with vendors. − PostgreSQL (BSD licence), for which Sun offers maintenance. Note that Oracle has also moved into the open source database segment by acquiring InnoDB (flagship engine included in MySQL) in 2005, to keep up with MySQL, and Sleepycat (publisher of BerkeleyDB, a database released under the BSD licence) in 2006. Oracle thus has a foot in the open source market and remains the company with the most resources to invest in it. In 2006, Oracle had discussed takeovers of JBoss and Zend. It is tricky to compare the economic weight of open source DBMS providers to that of proprietary software vendors based solely on sales: Oracle’s licensing revenues amounted to

Software - IT services

I 33

Sector report

$7bn in 2006 and those of MySQL to $40m, with the latter deriving exclusively from maintenance. Considering that there are 11m MySQL installations (company figure) out of a total market of 44m, MySQL has an estimated 25% of the database market. According to Ovum, MySQL was the leader in the open source DBMS segment in 2006 with estimated market share of 49%. MySQL is the best-known name worldwide in the database market, since it is included in the Linux LAMP (Linux, Apache, MySQL, PHP) stack. Chart 17: Open Source database Market presence Risky bets

Strong Contendors performers

Leaders Ingres

+ PostgreSQL

Offering

MySQL

Derby

Strategy

-

+

Source: Forrester Wave (2006)

Gartner reckons that open source DBMS are sufficiently mature to be used with non-critical applications. Where critical applications are concerned, IBM’s recent decision to bring MySQL to the System i for i5/OS was a first inroad, even if proprietary solutions (Oracle 11g, IBM’s DB2) offer better control of technologies like clustering. Clustering, or grid computing, makes databases more reliable by assigning several servers to manage a very large database. The strong growth seen in open source DBMS is being driven by: − Operating considerations: the advantage of open source databases over proprietary solutions is their time to market (continuous security upgrades). It thus suffices for database architects to migrate their entire bases to gain access to certain security patches that are only available with subsequent versions. Continuous updates help avoid shutting down information systems, and thus reduce operational risks. − The fact that the TCO on open source DBMS is 10 times lower over three years than for proprietary solutions, as mentioned above. Open source application servers With existing information systems, applications are involved in all business processes, from production through to accounting. Formerly used on different computers, these applications are increasingly being centralised on application servers that process all functions, allowing x

Software - IT services

I 34

Sector report

customers to use the same application (management of user sessions, load balancing, network connections and access to multiple data sources). Chart 18: System architecture OS: Solaris 10 11/06 (64-bit)

PostgreSQL JDBC TYPE IV 8.2

Client

HTTP Web Server

Application server

Javascript application web server (Servlets & JSP, EJB)

DBMS/R

PostgreSQL 8.2.4

Sun Java (TM) Systems Application Server 9.0, with elements --Servlet container: Jakarta Tomcat 5.5.16 --Java (TM) 2 Platform Standard Edition Development Kit 6.0 32-bit

Source: spec.org

Application server platforms (all technologies) form one of the largest segments of the middleware market, accounting for 18% of total revenue with licensing revenue of $2.1bn in 2006 according to Gartner. The segment could be worth as much as $5.2bn in 2009. It is divided between Java, .NET (Microsoft), PHP (for which Apache centralises developments) and Python technologies. Java offers a larger selection of vendors (IBM, Sun, Oracle, BEA Systems and open source). Solutions are differentiated by licence costs, the complexity of the servers, the level of administration required and the availability of technical support. IBM and BEA Systems lead the market for application platforms. BEA Systems started out as number one, but IBM surpassed it following the adoption of J2EE architecture. This development weakened BEA Systems enough to allow it to be taken over recently by Oracle.

Software - IT services

I 35

Sector report

Table 18: World market for application server platforms (pas seulement J2EE) $m

2005

2006 05 market share (%) 06 market share (%)

IBM BEA Systems Autres Hitachi Fujitsu Borland Oracle Adobe NEC Sun TmaxSoft Sybase Progress Software Pramati Kabira Red Hat (JBoss) Kingdee Software (China)

589.2 601.6 206.7 114.9 74.2 67.6 45.0 37.4 26.2 25.4 16.8 31.8 14.6 7.6 5.5 4.5 0.7

719.7 705.0 212.6 107.6 76.3 50.5 49.0 46.0 38.0 31.9 29.3 29.1 16.0 8.2 7.6 7.1 1.2

31.5 32.2 11.1 6.1 4.0 3.6 2.4 2.0 1.4 1.4 0.9 1.7 0.8 0.4 0.3 0.2 0.0

1 869.9 2 135.2

100.0

Total

Change 05 (%)

Change 06 (%)

33.7 33.0 10.0 5.0 3.6 2.4 2.3 2.2 1.8 1.5 1.4 1.4 0.7 0.4 0.4 0.3 0.1

15.0 11.3 -5.0 -4.6 -5.7 -14.1 50.8 8.9 -20.9 -43.5 35.0 3.0 65.5 14.1 6.8 28.6 121.3

22.2 17.2 2.8 -6.3 2.8 -25.4 8.9 23.2 45.1 25.5 74.4 -8.6 9.7 7.4 38.7 57.4 58.7

100.0

6.4

14.2

Source: Gartner (May 2007)

The fact that Gartner only takes into account licensing revenues puts open source suppliers at a disadvantage. Thus, JBoss only accounts for 0.3% of the market. To analyse the market in more depth, it is preferable to measure the installed base (the number of downloads alone does not tell whether the server is being used), and compare application servers depending on the technology (i.e. comparing J2EE servers to one another). BZ Research estimates that, in the J2EE server segment alone, JBoss had a market share of 32% in 2006, compared with 36.9% for IBM WebSphere. GlassFish v2, Sun’s open source application server, earned a spot in the ranking in 2006 with 4.5% market share and could become JBoss’s most serious competitor. The Geronimo/Apache application server is an entirely community-built project for which IBM offers maintenance via WebSphere Community Edition. Table 19: Penetration rate by technology in J2EE application servers (%) IBM WebSphere App Server JBoss Application Server BEA WebLogic Oracle App Serveur Sun Java Enterprise System Geronimo/Apache SAP Netweaver Macromedia/Adobe JRun GlassFish Borland AppServer Orion Server Sybase EAserver

2005 37.2 37.0 27.2 27.2 19.7 6.8 3.9 2.1 2.0

2006 36.9 32.0 23.7 22.4 19.0 11.8 6.0 5.6 4.5 3.2 2.1 1.9

Sources: BZresearch (2006) / % usage in polled companies

Software - IT services

I 36

Sector report

Forrester highlighted the low cost of maintenance services for the JBoss platform, which are lower than those for Sun and Oracle. It might be thought that with the tie-up between BEA Systems and Oracle, some clients might decide to switch to JBoss. Migrations from BEA Systems’ WebLogic towards Red Hat could then increase. Chart 19: J2EE application server platforms Market presence Risky bets

Contendors

Strong performers

+

Leaders

IBM Oracle Sun Offering

BEA JBOSS SAP

Strategy

-

+

Source: Forrester Wave (2006)

JBoss’ strength is also based on its SOA architecture, which was reinforced thanks to the acquisition of MetaMatrix, a vendor of a data integration and management solution. Open source IDEs Integrated development environments (IDEs) bring together programming and debugging of code. More specifically, IDEs make it possible to build graphic interfaces, vertical modules and Web applications. Among the most commonly used tools are IDEs for Java and PHP developments. The use of IDEs results from the popularity of programming languages. According to the August 2007 TIOBE Programming Community Index, Java dominates the programming language market with a 21.8% share.

Software - IT services

I 37

Sector report

Table 20: Worldwide market shares of programming languages Ranking August 07 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17

Programming language Java C (Visual) Basic C++ PHP Perl C# Python JavaScript Ruby PL/SQL SAS D Delphi Lua COBOL ABAP (SAP)

Market share August 07 (%) 21.8 15.7 10.6 10.1 9.7 5.3 4.0 2.7 2.6 1.9 1.8 1.4 1.2 1.2 0.6 0.6 0.6

Change (%) -0.6 -1.7 -0.5 -0.1 0.1 0.0 1.3 -0.3 0.6 1.1 0.7 -0.6 0.8 -0.9 0.6 0.0 0.2

Source: www.tiobe.com

The success of the Eclipse platform, an entirely open source IDE, is based on its widespread adoption by developers. According to BZ Research, in 2006 it had a market share of 70% in Java development (2.27 million users according to IDC in 2006). Faced with Eclipse, Sun has its own open source Java IDE: NetBeans. Open source Web languages (PHP, Python and Ruby) are enjoying even more rapid success than Java. Even though they do not have the same level of sophistication as Java, they are amply sufficient for most of the applications developing on the Web.

The applications layer: an offering gaining in maturity Numerous solutions are gaining in maturity in fields such as identity management (Open ID recently chosen by Orange for its 40 million subscribers), directories (openLDAP, open source implementation of the LDAP protocol), content management (Alfresco), BI (Jaspersoft, Pentaho) and ERP (Compiere, OpenBravo, TinyERP, SugarCRM, ERP5).

Software - IT services

I 38

Sector report

Table 21: Overview of open source applications Fields

Operating systems

Open source software Debian Mandriva RedHat Suse Novell Ubuntu

Directory

openLDAP

ERP

Compiere, OpenBravo, TinyERP, ERP5

CRM

SugarCRM

Business Intelligence

BIRT Jaspersoft (including Talend) Pentaho SpagoBI

Identity management

Open Trust, OpenID

Network management

Nagios

Electronic document management

Nuxeo Alfresco

Collaborative work

openXchange ExoPlatform

Wiki/Web 2.0 Virtualisation Geographic information systems Messaging Office suites

Xwiki Xen Grass Gis, MapServer OBM (Linagora), Jabber OpenOffice

Sources: Novaforge.org, Natixis Securities

GNU/Linux on work stations for companies The market in operating systems for work stations is still largely dominated by Microsoft. According to W3C figures, Windows XP was the world leader in operating systems for work stations in June 2007 with a market share of 74.6%. Windows 2000 was in second position with 6.2%, followed by Mac OS with 4%, and GNU/Linux, an open source solution, with 3.4% ahead of Vista at 3%. For its part, Gartner estimates the market share of Linux at 2.9% in 2007. Gartner expects a 2007/2011 CAGR of 10.5%, which would mean a market share of 3.7% in 2011. Linux could take advantage of the release of Vista to expand on the market in operating systems for work stations as Vista has not received the support of: − Manufacturers such as Toshiba, Acer, Asus, Lenovo (Lenovo announced that by the end of the year it planned to equip its Thinkpad portable computers with Novell’s SUSE Linux Enterprise Desktop 10) and Dell (Dell recently announced new machines sold without Windows and with the Linux Ubuntu distribution in the USA and Europe), which are starting to prefer to sell their machines with an GNU/Linux operating system that is less expensive than Windows, but with higher margins (no licence fees to be paid to Microsoft). − CIOs as Vista requires a lot of hardware resources and is accelerating the obsolescence of existing computers.

Software - IT services

I 39

Sector report

Ubuntu is the leading GNU/Linux (work stations and servers) distribution with a market share of 30% (around 15 million downloads) with Linux users worldwide in 2007. The Ubuntu distribution, developed by Canonical, was recently chosen to equip the work stations at the National Assembly in France. Chart 20: Ranking of the most frequently downloaded GNU/Linux distributions

Others Ubuntu

22.5%

29.2% Slackware 4.6% Mandriva 4.8% Fedora 7.0%

Debian Others 0.60%

12.2% openSUSE 10.1%

Source: Desktop Linux Market Survey (2007)

Canonical rapidly established itself with users, thanks to its simplified interface and its management based on graphic interfaces instead of commands. The distribution has also established itself with manufacturers such as Dell, HP, Lenovo and soon Acer. Mandriva is the only independent Linux vendor in Europe after the acquisition of Suse Linux by Novell. The company markets the Mandriva Linux operating system and has five million users. Following the acquisitions of Edge-IT (2004, France), Connectiva (2005, Brazil), Licoris (2005, Canada) and Linbox (2007, France), the company has 83 employees, including 25 in Brazil. Mandriva is also active in Poland, Russia, Morocco and the UK. Table 22: The Mandriva offering For individuals Subscription to the Powerpack

The subscription gives access to system updates and + 2 versions of Mandriva Linux Powerpack per year Mandriva Linux Powerpack 2008 Mandriva Linux Powerpack 2008 is the complete platform (server functions and middleware applications: SQL databases & Web frameworks + full Linux office + drivers – open or proprietary plugins) for all types of users. Mandriva flash Mandriva system in a USB key For companies Server solution Work stations Firewall

3-year maintenance 3-year maintenance NuFW

Price (€) 49 69 89 390 229 1,644

Source: Mandriva

All the products are derived from the Mandriva Linux project with a range for consumers and a range dedicated to companies. These products can be marketed on their own or as part of a broader service contract.

Software - IT services

I 40

Sector report

For the product sales activity, the role of Mandriva is to collect, compile, test and integrate components (a Linux solution comprises between 1,500 and 3,000 source packets) into a coherent product that can be used by companies and consumers. Mandriva is also developing an installation system (DrakX) as well as configuration tools allowing users to be immediately productive. Mandriva also maintains its code over time. On the Linux distributions market, Mandriva started up at the same time as Red Hat but suffered from a lack of cohesion in the user community. Whereas Red Hat divided the community part and the pay offering part for companies, to begin with Mandriva proposed the Mandriva Club, a pay subscription offering for the distribution, which entitled subscribers to certain privileges. The club apparently split the user community in two and tarnished Mandriva’s popularity. Next, Mandriva failed to propose a coherent product road map, for example by making available at the same time community version and the commercial version, and including superfluous services (E-learning). Mandriva’s vision of the market was too naive with an ill-defined target mixing beginners and experts, unlike Ubuntu, which directly targeted the beginners market. Furthermore, whereas Red Hat and Novell quickly focused on the server market and played on the Linux / Unix clash, Mandriva sought to compete with Microsoft. Last, its numerous acquisitions of services companies disrupted the turnaround of the business. Since September 2007 and the release of Mandriva Linux 2008 (>1 million downloads in the first week it was made available on P2P networks), the vendor has been preparing an overhaul of its offering for the community, with the shelving of its users Club. Mandriva is also simplifying its commercial range with a single PowerPack 2008 offering (previously there were three versions: discovery, hard user, SME) that can be configured according to the level of expertise. Mandriva also added its own tools for migrating from Microsoft Windows office suites to Mandriva Linux (Pullse) via a simplified interface. ERP open source: a threat to proprietary software vendors in the midmarket segment The ERP market is characterised by five trends: 1/ strong concentration, with 48% of the market being held by three software groups (SAP, Oracle, Microsoft) for a market worth a total of $36.7bn; 2/ mature generic technology; 3/ specialisation of ERP solutions by business; 4/ growing midmarket demand; and 5/ rapid growth of the SaaS model. According to the most recent Independent Oracle User Group Survey 2007, 33% of Oracle clients used an open source database and 13% run open source applications. 63% of Oracle clients say that they use open source software for simple functions (37% for departmental systems, 34% for websites and 12% for ERP and business intelligence). The youthful nature of open source ERP solutions is both a strength and a weakness vis-à-vis rival proprietary software. A strength because they have been developed with the most recent object languages and are highly flexible. A weakness because the tools do not have all the functions of proprietary solutions. For the time being, open source ERPs such as Compiere, OpenBravo or SugarCRM first reproduce the basic functions of proprietary ERP software packages: commercial management, procurement, stocks, production, human resources, sales force. They then develop the most sophisticated elements in different modules. Other initiatives, such as that of Apache OFbiz (open source suite bringing together ERP, CRM, SCM, e-commerce modules), which focus more on the development of the core software architecture, should bring about improvements in the capacities of open source ERPs. Software - IT services

I 41

Sector report

Table 23: Overview of open source ERPs ERP

Licence

Environment

Characteristics

Compiere

MPL

J2EE, JBOSS

The most commonly used ERP in the world, the most comprehensive (900,000 downloads).

ERP5

GPL

Python, Zope

ERP5 is optimised for industry and the public sector. It is available under the SaaS model like all the open source ERP solutions shown in the table.

TinyERP

GPL

Python

Very comprehensive, based on ERP5. Includes a BPM module. Its ORM (ObjectRelational Mapping) architecture, which is the most advanced, makes it possible to reduce the cost of the specific functions to be implemented (the most flexible).

OpenBravo

MPL

J2EE

OpenBravo recently broadened its offering with the acquisition of the open source project TinaPOS, a point-of-sale application designed for touch screens.

Opentaps

HPL

J2EE

Production version of the Apache OFBiz project.

SugarCRM

GPL

PHP

CRM solutions dedicated to the optimisation of commercial efforts. Solutions available ondemand and on-site.

Source: Natixis Securities

The commercial argument that there is no licence cost involved in open source ERP is invalid for large accounts as most of the costs of an ERP project relate to integration (75%) and the implementation time is longer with an open source solution. The main obstacle to the growth of open source ERPs is the costly investments already made in proprietary solutions. For CIOs, a full migration to open source solutions is hard to justify economically. However, the fact of not being tied to the vendor and being able to rework the code is arousing interest among major accounts. IDC puts the market share of open source ERP solutions at 7% in 2008 for an overall market that it estimates at €26.7bn. By our estimates, open source ERP vendors are likely to primarily target the small company market as licence costs are a bigger factor in the overall solution budget (46% vs. 23% for large accounts) and because ERP penetration has been weak so far.

Software - IT services

I 42

Sector report

Chart 21: Differing costs of support and maintenance licences for large accounts and the midmarket

Corporate

SMEs

23%

34%

27%

43%

36%

37%

46%

SoHo

43%

Licences

Support

11%

Maintenance

Source: IDC (2004)

The rapid growth of SaaS (software leasing on demand model managed by a hoster) in the midmarket segment and its adoption by open source software companies should favour the penetration of open source ERP solutions in this segment. The SaaS/open source combination meets the demand for low cost and flexibility of use. While the threat of open source ERP in large accounts seems weak over the next three years, we believe it should be taken seriously in the bottom end of the midmarket segment. At SAP, open source solutions may thus be direct rivals for Business One and, more importantly, the future SaaS Business ByDesign solution, the spearhead for the group’s growth in the medium term.

» ERP « Free

ADONIX

QAD (MFG/PRO)

ORACLE

NAVISION (MS)

TPE ($800m

SAP (Business Suite)

Chart 22: Segmentations of ERP offerings by category of clients

Sources: ENST, Natixis Securities

Software - IT services

I 43

Sector report

Business Intelligence also under attack Open source BI is still a long way from being a standard in IT systems despite a clear cost advantage (saving of around 30% on the total cost of a decision-making project). According to a 2006 Ventana study of IT personnel (architects, administrators, developers) in US organisations (sample of 320 replies), cost is the main criterion in decisions for just 16% of those surveyed. IT managers are still wary regarding the operating aspects of open source BI bricks (application security, administration tools, lack of support on the part of software groups), especially in the implementation of major projects (1,000 users). As a result, 79% of those surveyed accord importance to support and do not want to end up with open source technologies that they have to manipulate themselves. Moreover, open source solutions have not attained the functional depth, robustness and availability of proprietary solutions. The survey nevertheless pointed to real momentum since 83% of those surveyed have already deployed, or are planning to deploy, open source BI bricks in their IT systems, especially for small projects (200 users) for 79% of respondents. The study also indicated that companies that have already invested in proprietary BI solutions will not be prompted to change their BI bricks because of the migration costs (particularly training costs, which account for 20% of the total budget for migration to non-proprietary solutions, according to Gartner, with assessment and portage of specific applications accounting for the other 80%). These migration costs also occur when switching from one proprietary solution to another. By virtue of their model, open source software groups cannot propose product credits or services comparable to those offered by SAP to Oracle clients. Open source software companies (such as Pentaho and JasperSoft), proprietary software groups (such as Actuate) and integrators (like SpagoBI) propose comprehensive suites including ETL, OLAP drivers and reports. In the value chain, they are active mainly in BI exploitation but also in the ETL segment via OEM agreements. Table 24: Software groups in the decision-making business chain Transactional

Database

ETL 1

Data warehouse

BI exploitation

Generate heterogeneous data

Stock data in real time

Process data

Archive data for exploitation

Exploit coherent data

Proprietary

SAP, Siebel (Oracle)

Oracle, DB2

IBM, Informatica

Teradata, Oracle

BO, Cognos

Open Source

SugarCRM

MySQL

Talend

Greenplum

Jaspersoft, Pentaho

1

ETL: extract, transform and load

Source: Natixis Securities

Jaspersoft, a company created in 2001, stands out on the open source BI segment, with a market share estimated at 50% (number of downloads), i.e. more than 3 million downloads from 82 countries for v1.0, compared with 3 million for Pentaho and 2 million for BIRT (which is not entirely open source). Jaspersoft proposes a packaged offering, which facilitates integration needs, and includes in its suite (Jaspersoft Business Intelligence Suite) ETL bricks (Talend integrated as original equipment in Jasper ETL), ad hoc query bricks (Jasper Server), reporting (Jasper Reports) and an OLAP engine (Jasper Analysis). Recently, Jaspersoft developed an offering that is interoperable with Salesforce software (Jasper4Salesforce),

Software - IT services

I 44

Sector report

CRM applications on demand. Other OEM partnerships were also forged with Ingres and MySQL, as well as with Red Hat’s RHS exchange platform, and more recently Jasper4Oracle E-Business Suite. Via its JBIS 2.0 solution, Jaspersoft could therefore benefit from a solid base to boost its presence vis-à-vis proprietary software makers, especially since the launch by Ingres of its Icebreaker BI Appliance decision-making offering, which brings together an Ingres database with a Linux Rpath operating system and Jaspersoft’s decision-making suite. Table 25: Overview of the leading open source BI offerings Solutions BIRT

Vendors Actuate

Type Decisionmaking suite

Jasper intelligence v2.0

Jaspersoft

Decisionmaking suite

SpagoBI v1.9.3

Engineering Ingegneria Informatica (EII)

Decisionmaking suite

Pentaho

Pentaho

Decisionmaking suite

Greenplum database, Bizgres

Greenplum

Data warehouse

Strong points Comprehensive decision-making suite (reporting, OLAP), IDE Eclipse plug-in for creation of reports. Comprehensive decision-making suite (reporting, analysis, ETL) based on JFreeReport for reports and on Talend for ETL.

Licence EPL (Eclipse Public Licence)

Comprehensive decision-making suite (reporting, OLAP), includes the best open source components (Jasper, Birt, Mondrian, Weka), and also BO, Microsoft Analysis Services proprietary components. Comprehensive decision-making suite (reporting, analysis, dashboard, data mining), based on JFreeReport for reports, Weka for datamining, Kettle for ETL and Mondrian for the OLAP server. G3 is the most recent version of its data warehouse that can handle large data volumes and process them rapidly.

LGPL

GPLv3

MPL (Mozilla Public License)

BSD (Berkeley Software Distribution)

Sources: Natixis Securities, companies

As part of this study, we met the management of Talend (2007 sales of $1m est., 70 employees), a French company specialised in the integration of decision-making data. Talend is a perfect illustration of the open-source business model. The data integration market is dominated by IBM (DataStage, technology acquired since the takeover of Ascential), with a market share of 31% in 2006. US pure player Informatica (PowerCenter 8.1) lies in second place with a market share of 22.5%. In data integration, Talend is positioned in the operational integration niche (infrastructure layer) and less in the decision-making integration sector (overall decision-making platform), where the big proprietary software groups are already very present. Operational integration consists in collecting and integrating data (for example ERP or CRM production bases under PostgreSQL, functioning with a Teradata or Greenplum data warehouse). Talend is specialised in database migrations, historical database loading, and synchronisation using open source connectors. This open source segment has growth potential as the IT systems of large accounts already have historical applications and infrastructure layers, and these groups are hesitant to invest further in solutions that do not have common standards. They thus turn to open source, which favours interoperability. According to Talend, the proprietary model is a curb on the ETL layer. The company’s CEO describes proprietary solutions as black boxes with a high entry fee (the licence), to which must be added the cost of buying connectors to link the databases to the warehouses. These connectors are very expensive, especially if the historical layer of the IT system comprises a

Software - IT services

I 45

Sector report

mix of proprietary technologies. Thus, in total, the budget for an Informatica solution can amount to $150k per production server for the licence plus $40k per year for maintenance. Historical proprietary software makers (Informatica/PowerCenter and AscentialIBM/DataStage) do not offer many connectors. Talend has 200 connectors covering several types of files, the main ERP and CRM systems, databases (MySQL, PostgreSQL, Oracle, Sybase, DB2, Firebird, Informix, Ingres, MS SQL, Java DB), MOMs (Middleware Oriented Messages) and Web services. Talend offers a GPLv2.0 community open source version of its integration suite, and a modified GPL open source version (the project code cannot be reproduced but the source code is available) for which users have to pay. Table 26: Talend’s platform offering Talend offering Talend Open Studio Talend Integration Suite Team Professional Enterprise Talend4SugarCRM Talend on demand

Versions for five users Community version GPLv2.0 Version with support services, maintenance and add-ons VSB target SME target Targets large accounts Integration with SugarCRM SaaS version (free for up to 3 users)

Price/year ($) Free 15,000 45,000 80,000 est. 4,995 199/month

Source: Talend

Talend disseminates its solutions via two distribution channels, directly via www.talend.com, which avoids the need to create a network of resellers, and indirectly via OEM partnerships: − With software companies such as: Jaspersoft (leading vendor of open source BI software with 5,000 clients, which includes JasperETL based on the Talend code generator), SugarCRM, Ingres (BI appliance with Jaspersoft), SpagoBI (which integrates JasperReport reports and therefore Talend software), and MySQL. More recently, Talend announced a partnership with Camptocamp in order to launch Spatial Data Integrator powered by Talend (a data integration solution specialised in the manipulation of geographical information). As part of OEM contracts, Talend is paid a percentage, on a case by case basis of the project implemented (