If your commercially supported Open Source software follows the high-level guidelines we defined in our original article, feel free to use it and link to cosmo.itredux.com. We’re also very interested to hear about your feedback regarding these guidelines, and how they could be improved upon.

Edition Layering
Past and current evidence suggests that no software company has ever been successful at generating the kind of gross margins that traditional software companies can produce, over an extended period of time (more than 5 years), while releasing 100% of their code under an OSI-approved license. A broadly accepted answer to this challenge is layering, whereby a subset of the product is available for free and/or under an Open Source license, while the complete product can only be used in the context of a commercial relationship with the vendor.

Dual or Triple Licensing
With a two-tier layering approach, the Open Source Edition is usually released under a viral license (GPL, LGPL, or similar licenses), and the Commercial Edition is released under a traditional End User License Agreement (dual licensing), while adding features and services to the Open Source Edition. With a three-tier layering approach like Intalio’s, an Open Source Edition is available under a liberal license (Apache, BSD, or MIT, etc.) and covers at least 80% of the overall product’s codebase, a Community Edition is available free of charge (for development and production), with or without source code, and a Commercial Edition (called Enterprise Edition in Intalio’s case) is made available under an End User License Agreement, while providing access to 100% of the source code, and granting customers the right to modify the original source code, without having to give modifications back to the vendor.

Open Source Foundation
With either two-tier or three-tier layering, over 80% of the overall code base is available under a pure Open Source license, as defined by the Open Source Initiative. When using a three-tier layering model, the middle-tier edition (Community Edition) might or might not be available under an Open Source license, but must be available free of charge, for both development and production usage.

Source Code Availability
100% of the Commercial Edition’s source code is made available to customers, with rights to modify the source code, and no obligations to submit modifications back to the vendor. Modifications made by customers to the original source code might not be supported by the vendor. If modifications are donated back to the vendor for inclusion into the original product, they must be accompanied with full transfer of intellectual property rights, usually granted through a signed contributor agreement.

Subscription-based Licensing
Unlike traditional software sold through perpetual licenses and yearly support contracts, the COSMO model is based on yearly subscriptions that include both licenses for the product’s Commercial Edition and support services. The later usually include patch updates for software upgrades, access to better documentation and a customer support infrastructure, etc. As a result, a product that is commercialized through the COSMO model is purchased from an operating expenses budget (OPEX), rather than a capital expenditure one (CAPEX).

A more detailed introduction to COSMO can be found on the Intalio 2.0 article.

There are two primary reasons why I like Amazon Web Services: One, they define very simple interfaces for addressing very common needs you have when developing any kind of Web-based application. Two, they promote a no non-sense pricing model that makes them a viable option for most projects. Thing is, I am not the only one thinking that way, and I am hearing more and more organizations, small and large, starting to adopt Amazon’s services, which raises a very important question: can we rely on them?

As any utlity, Amazon Web Services are subject to failure, and recent outages experienced by developers are a painful reminder than when you put all your eggs in the same basket, and the basket falls on the ground, you better like your eggs scrambled. Which is why you should use multiple baskets.

Imagine for a moment that an open-source project would be created with the goal of developing all the software necessary for deploying clones of Amazon’s most popular Web services, including EC2, S3, SQS. Imagine that you could deploy this open-source software onto your own servers. Imagine that more elaborate versions would be developed on top of some kind of grid operating system — like 3TERA, allowing any ISP or Web hosting provider to provide plug-compatible Amazon-like services by using unused bandwidth, processing, and storage capacity. Finally, imagine that such an open-source project would also develop the client-side software you need to use Amazon’s Web services in combination with any other clone in order to get failover capabilities, without having to learn any new API. Do you think such a thing would be of value, to you or others?

I recently asked this very question to a group of five IT executives working for one of the ten largest companies in the world, and the answer was a resounding ‘yes’. I must assume that you could find many other companies, small and large, that would share the same opinion.

If such a thing existed, you would be able to use Amazon’s low cost services for your daily operations, and the services offered by more expensive providers whenever Amazon would go down, for some limited period of time. You would also be able to run your own version of Amazon’s services within your own data center as a backup, giving you the assurance that your data and your services remain available, even in the event of a catastrophic failure, or a decision to increase prices beyond reason.

Today, popular Web sites and mission-critical applications served over the Web rely on redundant connections to the Internet that are provided by multiple service providers, as a way to remove any single point of failure. Time has come to apply the same approach a level up in the stack, and I could not think of a better starting point than Amazon Web Services for it.

Initially, Amazon might see such a project as a competitive threat, but over time, it should realize that most developers will still use its original services as primary ones, for they are most likely to provide the highest level of availability, at the lowest price point.

Such an open-source project does not exist today, but I bet it will very soon.

Creating a mutually-beneficial partnership is a difficult exercise. In order to make it work, each party must get something from the relationship that it could not get on its own, otherwise it is clearly leaving something on the table. Also, it must ensure that the lure of a short term gain does not come at the expense of a long term loss. For these reasons, most partnerships established between enterprise software companies tend to be asymmetrical, for symmetrical ones tend to create unbearable competitive situations or unbalanced risk/reward ratios that would undermine the partnership over time.

Enterprise software is all about customer ownership, and two large vendors usually find it difficult to share customer relationships on the long term. As a result, it’s a lot easier for a large vendor to source missing pieces of technology from a smaller vendor that does not have a significant direct sales force, and could easily be acquired down the road.

Similarly, System Integrators and Value Added Resellers adding a new technology to their portfolio make a risky bet, investing significant resources in terms of business development and training, without much assurance that customers will necessarily have a need for the technology, or will be comfortable dealing with its provider. For this reason, SIs and VARs tend to focus their efforts on mature technologies backed by established players, which is one of the reasons why vendors like IBM and Microsoft have so many SI and VAR partners.

The emergence of Commercial Open Source vendors like Alfresco, MySQL, or SugarCRM is helping break the status quo. These vendors tend to have complementary solutions that customers like to integrate within larger stacks, therefore relieving some competitive pressure. Also, Commercial Open Source business models rely heavily on user adoption and customer conversion rates. As a result, the more users get to adopt your product, and the faster they convert into paying customers for your subscription services, the better. And if they happen to subscribe to the services of four or five vendors providing bits and pieces of the Open Source stack they are building, everybody wins.

Another dynamic is helping turn this stack centric partnership process into a virtuous circle: when a customer has gone through the process of evaluating and adopting a piece of Commercial Open Source technology, while developing detailed Return on Investment models that strongly will support its selection of an Open Source vendor over any closed-source alternative, it is much more likely to seriously consider another Commercial Open Source provider for any new need that might emerge down the road.

The benefits of such a process have certainly not escaped the new crop of Commercial Open Source vendors, and some interesting stacks are starting to emerge, adopting a very simple business model that customers seem to be comfortable with. According to this model, one vendor embeds the Open Source versions of the products offered by several other Commercial Open Source vendors, while establishing reselling or OEM agreements with them for their commercial offerings — usually refered to as Enterprise Editions, giving customers the ability to upgrade on a component per component basis. Such a model offers the benefit of significantly reducing the long term risk for the vendor responsible for the integration of a complete stack, for the reason that all the pieces that are integrated are available under the terms of an Open Source license. At the same time, it creates additional distribution channels for the vendors that are part of the stack, in exchange for very minimal investments.

An example of such a stack is the one currently developed by Intalio around our Business Process Management System (BPMS). Over time, we realized that our customers needed more than just a process design tool, a process execution engine, and a workflow framework. Some wanted integration with a modern Enterprise Content Management (ECM) system in order to support scenarios where BPM and ECM intersect, others wanted to get the functionality offered by a complete Enterprise Service Bus (ESB) as a way to build their own SOA puzzle.

At present time, Intalio is working with the following vendors to build a process-centric stack: MuleSource for the ESB, OpenLexicon for business rules, Liferay for the portal, Orbeon for dynamic forms, Alfresco for ECM, and Hyperic for system management. Interestingly enough, larger vendors are starting to adopt this model as well, and we recently signed a partnership with IBM for WebSphere Application Server Community Edition (WAS CE), following a very similar model. Down the road, we expect more partnerships of this nature to be established, and our relationships with these vendors to become even closer, with reselling agreements turning into OEM agreements when integration points have demonstrated to bring value to customers that can be converted into additional subscription services.

Here is yet another benefit offered by a Commercial Open Source model.

First, a little bit of history. Intalio was started in 1999 as a pure Open Source venture. 100% of the code we produced was licensed under an Apache-like license, which is as liberal as it gets beyond public domain. We developed projects such as OpenEJB — which became part of Apache Geronimo, OpenJMS, OpenORB, Castor, Tyrex, and OpenXML, which was merged with IBM’s XML4J to serve as a foundation for Apache Xerces. We then built the first-generation of our standards-based Business Process Management System (Intalio|n³) on top of some of these Open Source components, using a traditional, closed-source licensing model. Eventually, we decided to come back to an Open Source model for our core product, and made an acquisition back in December 2005 in order to accelerate our transition from one standard process execution language (BPML) to another (BPEL). We announced our new Open Source strategy back then, and made a first release of our second-generation product in February 2006.

For many reasons, we decided to make this transition to Open Source a process, rather than a singular event. First, we had to get rid of third-party components that we had used to build our product, and that were not available under an Open Source license, such as the library we used to draw BPMN diagrams. Second, we wanted to leverage existing Open Source communities, such as the Eclipse Foundation or the Apache Software Foundation, rather than building a new community from scratch, therefore had to follow their own processes, like the Apache Incubator process. Third, having been in the Open Source game for over six years, we knew very well that we would have to use a mix of layering and dual licensing to make our business sustainable, and we did not want to mislead our contributors, customers, and partners by going 100% Open Source initially, then adding non Open Source components to our business over time. Trust matters if you want your business to last.

So let’s take a look at our track record: First, we donated our BPEL engine to Apache back in February, making it the core codebase for the Apache Ode project. Second, we donated our BPMN modeler to Eclipse. Third, we released our workflow framework under the Eclipse Public License, and made it available at tempo.intalio.org. Fourth, we announced that we will release our entire BPMS under the Mozilla Public License (MPL) amended with an attribution clause. This full Open Source release will be made available at intalio.org sometime during the current quarter, and we hope that reasonable attribution terms will be approved by the Open Source Initiative (OSI) by then.

When we make this release, two editions of our product will be made available: Intalio|BPMS Community Edition and Intalio|BPMS Enterprise Edition. The Community Edition will be available under the terms of an amended MPL, but we will not provide any technical support for it. The Enterprise Edition will be available under a traditional end user licensing agreement, and we will provide technical support for it through yearly subscription contracts, giving customers access to patch updates and some additional features, following the exact same model as JBoss for the JBoss Operations Network. The Community Edition will share approximately 95% of the code found in the Enterprise Edition, and we expect this ratio to remain constant over time. This is our commitment to an Open Source model that we believe will benefit our customers, our investors, and the community at large.

That being said, there is a lot more to Commercial Open Source than just licensing terms. To me, the implications of an Open Source business model are a lot more significant when considering the inverted sales process it can support. The enterprise software industry needs to be reformed in many ways, and providing source code under an Open Source license is only a small part of the changes that are required. Allowing enterprise customers to adopt radically new technologies while lowering all barriers to adoption is what CIOs should be looking for, and the inverted sales process pioneered by earlier commercial Open Source players — il faut rendre à César ce qui appartient à César — is what it’s all about.

To conclude, if the simple idea of having a complete BPMS available under an Open Source license is appealing to you, I invite you to contribute to Intalio’s various Open Source projects. You might find that adding a BPMN modeler to your BPM product would offer some benefits to your customers. You might also realize that supporting BPEL will be a good thing moving forward. Or you might want to reuse some pieces of our BPEL4People workflow framework. In all cases, the Open Source licenses we used for these different projects will allow you to do so without having to buy a license from us, which is nice in and by itself. And if you find some pieces missing, join us and let’s build them together.

“It’s always more fun to share with everyone.”
-Jack Johnson

Reference: Intalio’s Contributions to Open Source

Open Source will drive BPM adoption
The release of a complete BPMS under an Open Source license should significantly reduce any barrier to adoption. The leading Open Source BPM vendor will capture the largest market share, while more companies will start BPM projects than in the prior three years combined.

Google will acquire a complete Office 2.0 suite
Google Docs is like ThinkFree’s Quick Edit mode: it works, but the feature set is way too limited to convince a large-enough number of Microsoft Office users to make the switch. Something akin to ThinkFree’s Power Edit mode is necessary, and buying it will certainly be quicker than building it. Google, which doubled its workforce to almost 10,000 employees in 2006 would rather spend cash and stock than waste time. ThinkFree or Zoho look like perfect targets, even though convincing the good folks at AdventNet won’t be an easy task. Apple Computer might make a similar move too, significantly enhancing the capabilities offered by the excellent .Mac online service. On the mobility front, Palm will likely be acquired as well, possibly by a telecommunication service provider.

Office 2.0 will start getting adoption in the corporate world
While early adopters will continue to be very small businesses and academic users, the corporate world will start playing with the technology as well. Individual users will favor large collections of applications such as Zoho, while corporate departments will prefer feature-rich suites such as ThinkFree. This movement will be acknowledged by several Gartner reports, as well as a cover article in BusinessWeek magazine.

All top ten off-shore software development firms will have a BPM practice
The Capability Maturity Model for Software (also known as CMM) has been retired, and the Capability Maturity Model Integration (CMMI) replaced it, with a stronger emphasis on business processes. The model-driven development approach enabled by Business Process Management (BPM) technologies makes it easier to measure and improve the quality of development processes compared to traditional software development approaches based on the writing of code. Off-shore software development firms have known that for a long time — Infosys was one of the first companies in the world to develop a working implementation of the BPML language back in 2001 — and will take advantage of it, now that commercial implementations have matured enough.

Open Source vendors will grow faster than their closed source competitors
More user accounts will be created on SugarCRM and vtiger combined than on Salesforce.com. Compiere will announce the successful migrations of several large Oracle E-Business Suite customers. Intalio will sign more new customers than Lombardi, Pegasystem, and Savvion combined. The vast majority of SOA projects will be using Open Source solutions such as Axis, Mule, and ServiceMix. And the first Open Source database vendor (EnterpriseDB, Ingres, or MySQL) to release a plug-compatible replacement for the Oracle database that can support the SAP R/3 application for over 10,000 concurrent users will get the best home run in database history since Sybase, but that might take a couple more years…

Apple Computer will make online data synchronization really work
Once the iPhone — or PodPhone, or whatever Steve decides to call it — is released, people will realize that its killer feature is not that it can play tunes, but that it can synchronize contacts and events in a wireless fashion more easily than any other system. Granted, such synchronization has been working fairly well with the BlackBerry+Exchange combination for quite sometime now, but who really likes to administrate a Microsoft Exchange server? Also, a Mac-compatible version of QuickBooks Online Edition will be released, unless a competitive offering steals the show first.

SOA will get simpler
The Java Business Integration (JBI) API will go the way of CORBA and be relegated to specific vertical industries such as telecommunication services. Large corporate users demanding very high transaction volumes — think Wall Street brokers & dealers — will favor lightweight application servers with streamlined stacks for Web services such as Apache Axis, and MuleSource will emerge as the leading Open Source ESB player. Production-grade ESB functionality will come for free with all major Open Source stacks, accelerating the demise of the few propriatery EAI vendors left standing.

Enterprise 2.0 will establish itself as a sticking meme
High-profile conferences originally focused on collaboration and e-Business will adopt the moniker, Oracle and SAP will incorporate Enterprise 2.0 stacks similar to Intel’s SuiteTwo within their respective product lines, and the SEC will issue new guidelines for corporate blogging.

Standards for online services will be adopted
Enough services supporting OpenID and Amazon’s Simple Storage Service (S3) will become available for early adopters to build a working Office 2.0 Setup that provides single sign-on and remote data storage. Open Source frameworks offering plug-compatible alternatives to Amazon’s EC2, S3, SQS will be released as well, enabling truly distributed backup strategies to be developed.

That’s it for now. Happy new year to you all again!

Disclaimer: I work for Intalio and serve as an advisor for ThinkFree and Zoho.

BPMS will go mainstream: Draw
The concept for a Business Process Management System (BPMS) has been gaining momentum, but I would not call it mainstream yet. The ground for BPM 2.0 has been laid, and the first Open Source BPMS has been announced. Nevertheless, customers are still trying to figure out what BPM really is, and which flavor for it will best address their needs initially. Gartner recently came to the rescue, suggesting that an Open Source solution might be a good way to get your feet wet initially. But a commonly-accepted compeling event for the deployment of a BPM platform — think e-Commerce for an application server — remains to be found. SOA might very well be it, but it’s losing ground as well, as illustrated on this Google Trends analysis, also mentioned in this earlier post. Like it or not, BPM is a complex concept, and it will take time for corporate buyers to get their arms around it. In the meantime, the best way to get there is to lower any possible barrier to adoption, and this is what Intalio is all about.

Industry consolidation will accelerate: Miss
No game-changing mergers or acquisitions were made in the enterprise software space this year, and IBM did not come back to the enterprise application game. Smaller deals have been stealing the show, especially Red Hat’s acquisition of JBoss, which must have compelled Oracle to finally make a move on the Linux front. The real question now comes down to this: how long will Salesforce.com remain as an independent company? IBM, Oracle, and SAP would all make pretty good buyers. Another option would be for Marc to finally extend his platform with ERP functionality, which could be achieved in several ways, one of them being to acquire Compiere or Intacct. Let’s hope the later comes true.

Office 2.0 will become a reality: Hit
Within less than nine months, the Office 2.0 Database, announced on this post, grew from nothing to over 330 applications. The concept introduced exactly a year ago got its first conference, with over 450 participants, 105 speakers, and 56 sponsors. Zoho emerged as a likely winner in the space, and ThinkFree demonstrated that web browsers are now powerful enough to support most of the features that are offered by traditional Office 1.0 productivity suites today [Disclaimer: I serve as an advisor for both companies]. Google’s release of
Google Docs on the first day of the Office 2.0 Conference also gave a clear signal that things are getting serious now, and that past involvements with OpenOffice.org were just smoke and mirrors. Next year, it will be interesting to see wether Microsoft finally releases something of substance under the Office Live brand, and who gets their hands on ThinkFree. The later would be a perfect acquisition for Google, but in the meantime an OEM deal with Intuit or Salesforce.com would make a lot of sense as well. Investors might also be interested to know that the company is giving serious thoughts to the idea of raising a small round of venture funding…

Software off-shoring will go further West: Hit
If your Indian off-shoring partner made you lose sleep over high employee turn-over, you might have considered going further West. During the past year, Eastern European countries such as Ukraine became a popular destination for the off-shoring of software development projects, so much so that the unescapable law of supply and demand kicked back in full swing. And because these countries are so much smaller than India, its effects were felt stronger, faster. Down the road, it seems that off-shoring is a dynamic process rather than a singular event, for which target destinations keep changing as local markets mature. Next steps? Latin America, Northern Africa, and China of course.

Open Source business models will prove themselves: Hit
Following on the footsteps of JBoss and SugarCRM, more and more enterprise software companies went the Open Source way, including Hyperic, Intalio, and Terracotta. Learning valuable lessons from their predecessors, many commercial Open Source upstarts are being liberal with licensing terms, which compelled some industry observers to challenge their claims. Next year should see a clarification of the rules, with the addition of generally agreed-upon terms for attribution to the ever more popular Mozilla Public License (MPL), with the blessing of the Open Source Initiative (OSI) hopefully.

All software will go the service way: Hit
Alongside the emergence of credible Office 2.0 solutions, all categories of software saw the release of online alternatives, including drawing, music playing, and video editing. A year ago, guessing the next category of software to go online was challenging. Today, the game has changed, and the challenge is to identify the one that will be the last. At this point, only one application is missing from my Office 2.0 Setup: a good online accounting tool.

SOA will get more complex: Hit
Service Oriented Architecture (SOA) was supposed to make integration simpler. Instead, it made everything a lot more complex, and the introduction of the Enterprise Service Bus (ESB) did not really help. Standards for Web Services kept multiplying, and the market kept fragmenting itself, with the Java Business Integration (JBI) API on one side, and slightly more pragmatic approaches such as Apache Axis on the other. Too many options create confusion, which vendors of proprietary solutions are the only ones to benefit from. Clarity and simplicity are much needed, but I’m not sure where they’ll come from next year.

Corporations will discover the beauty of social: Hit
LinkedIn will most likely sign its tenth million users in the coming quarter, and TIME Magazine named You (and that includes Me) as Person of the Year. Most searches made on Google return entries from Wikipedia within the top ten results, while the Internet Movie Database (IMDB) usually tops official websites developed by major studios for any new movie. During this past year, the corporate world started to pay attention, and the concept for Enterprise 2.0 was born, followed by commercial implementations such as Intel’s SuiteTwo. Next year should see an acceleration of the trend, with more and more practical use cases and success stories being released.

Standards for online services will emerge: Hit
Bugs have been added to the Office 2.0 Bug Tracker faster than standards have been released to fix them, nevertheless, a couple of initiatives have been gaining some significant traction, among them JSON and OpenID. Also, and against all expectations, Amazon started to release a set of commodity services (EC2, S3, SQS) for online computing that’s giving us a very good preview of things to come. Next year should see Google and Microsoft following on Jeff’s footsteps. The creation of a full fledge standardization body for Office 2.0 is not out of the question either.

Score: 7 Hits, 1 Miss, 1 Draw
Let’s call it a 83% success rate. Not too bad… In fact, it’s a little bit too good, and the reason for it might be that my inferences for ‘06 were a little bit too open ended. For next year’s, I will try to be a little bit more specific, which should make the challenge more interesting. See you next year!

I am in Las Vegas for SAP TechEd with Michael Cote, Mark Crofton, Tim Marman, Mike Masnick, Niel Robertson, and Charlie Wood. Back in May, SAP was one of the first IT vendors to invite bloggers to cover its user conference, SAPPHIRE. Today, they did it again, this time around for their developer event.

Back then, I wrote about SAP’s Business Process Platform, explained why SAP should Open Source NetWeaver, and advocated for duos in complement to Duet. Today, through Shai Agassi’s keynote and a couple of executive briefings, I got a nice update on what SAP is up to these days. Before getting into details, all I can say is that it’s not your father’s SAP anymore. The market for enterprise software is undergoing one of the major transformations it’s going through every ten years or so, and SAP is embracing these changes better than many other vendors I could think of. Here is what SAP is getting right, and here is where it could do an even better job.

On the technology side, the Business Process Platform is coming together. It does not have the advanced process orchestration capabilities that I would like to see at this level of the stack, but the composition tools that were showcased during Shai’s keynote sure look like a step in the right direction. The Business Intelligence Accelerator is also very impressive. If BI can be made sexy, BAI certainly is one way to do it. The user interface is extremely slick — I love the spaghetti effect when refreshing charts — and performance improvements of up to 100x sure look good.

On the user interface side, Muse is one of the sexiest user interfaces I’ve seen for any large enterprise application, and the folks from SAP Labs in Palo Alto keep innovating with some sweet eye candies. Learning a couple of tricks from Apple, they took Konfabulator, now owned by Yahoo!, and built a really cool workflow task list that can be displayed on any desktop. They also added a voice activated interface for it, demonstrating that getting all the data in one place and providing multiple interfaces for it is possible, but most importantly useful. Now, being the Office 2.0 biggot that I am, I’d love to get a version of this that I could use as a NetVibes portlet, or Google Homepage component, without having to install anything on my computer. Note to Eric Wood: I love your widgets, give us more of them!

On the partnership side, SAP is changing too. It used to be difficult for an independent software vendor based outside of Germany to develop a meaningful partnership with SAP. And a strong case of not-invented-here syndrome made it virtually impossible for SAP to bring innovation in through acquisitions. This is changing. The NetWeaver ecosystem is now strong of more than 400 vendors, and successful acquisitions of companies such as Virsa are clear evidence that SAP opened up to the outside world. This ecosystem is nowhere near as strong as what IBM has been nurturing over the past fifteen years or so, but it’s growing fast. There are a couple of things that SAP could do to take it to the next level, and I’ll write about it tomorrow.

On the community side, SDN has over 500,000 registered users today. It makes it one of the largest communities of developers in the world. Not yet on par with Microsoft’s 8M Visual developers, or Sun’s 3 million Java folks, but definitely comparable to Oracle’s community of PL/SQL experts. Now, this group of half a million might overlap quite significantly with the half million folks who have been trained to ABAP, therefore SAP must go the extra mile to reach out beyond its existing base. One initiative of interest to me is the work on scripting languages promoted by folks like Craig Cmehil, which is a good way to engage with a broader community of developers who might prefer PHP to ABAP — not that I fully understand why anyone would…

On the product side, the biggest announcement to come out of TechEd is the one on mySAP ERP 2005, which is now presented as the stable core that SAP customers should migrate to, from their existing R/3 4.6c or 4.7 deployments. SAP is committing to keep this core stable for the next five years, and to release two to three enhancement packages every year. Setting a clear update target is a smart move, which seems to be coming right on time. According to Peter Graf, more than half of SAP’s customers running R/3 4.6c or 4.7 today should have upgraded to mySAP ERP 2005 by 2008. This seems realistic, and SAP is doing a pretty good job building a compeling business case for it. Migration tools have been significantly improved as well, and the experience of early adopters that have done this upgrade already should help. This won’t make it significantly easier for the very largest SAP customers that are running hundreds of heavily-customized instances of R/3 today, and this is where BPM 2.0 could provide part of the answer. Note to the Business Process Platform folks: feel free to call us, we’d love to help.

Finally, Shai Agassi clarified a lot of issues related to SAP’s contribution to the Open Source community. Taking the LAMP stack as an example, he reminded our group of analysts and bloggers that SAP R/3 was the first ERP to be certified for Linux, and that SAP Ventures invested in both MySQL and Zend, original developers for the massively successful PHP scripting language. Arguably, SAP is one of the large IT vendors that is making the most significant contributions to the Open Source community today, but it could — and should — do more. As Shai made it crystal clear, SAP is neither a platform company, nor a technology company, it’s an application company. In his own words, it’s the company that can best “codify business processes into applications.” I wholeheartedly agree with him, and would push the thinking a step further: if SAP is an application company, then SAP should open the underlying platform — NetWeaver — to a broader audience. As advocated before, SAP should make NetWeaver available under an Open Source license, or at least release some Community Edition for it under very favorable licensing terms for ISVs, much like IBM did with WebSphere. I will write about this some more tomorrow.

Overall, I think that SAP is doing a phenomenal job of opening up to the world. Its support for enterprise bloggers will most likely turn into a use case for how corporations can foster conversations with the community at large, while preserving the integrity of all parties involved. My only piece of advice there is pretty simple: keep doing what you’re doing today.

At this point, I can only look forward to SAPPHIRE ‘07 in Atlanta.

Why don’t analysts cover Intalio more?
Because we do not pay them anymore. Sorry, I could not resist… Yet, this is not far from the truth. We do not spend enough time with them, for we have very limited marketing resources. I am the one responsible for this, and I am trying to fix it, with an analyst roadshow to take place in September. During my seven years working for Intalio, I’ve learned that industry analysts won’t say much about you if you do not brief them first, in a proactive manner. Analysts from the private banking world do though. In fact, the best research seems to be made by bankers, which sort of makes sense when you think about it: what they research has a direct impact on their own money, not their customers’ money… Overall, I think analysts should be paid by customers, not vendors. This would add a lot more credibility to their work. But who am I to say that?

Will we integrate Intalio|BPMS with an Enterprise Portal?
It’s done already. the user interface components of Intalio|Workflow are already packaged as JSR 168 components. And we have a customer in Canada who is working on some integration with Liferay Enterprise Portal, which James referred to. We are also considering integration with content management systems such as Alfresco for example.

What should the Open Source community do to promote our sales process?
CEOs of Open Source companies already do that to a certain extent, and Marc Fleury is a good example for this. What’s missing is more metrics on what’s working, and what’s not, and clarification around the concept of downloads. Part of the problem is that the number of downloads seems to be the metric that Open Source companies are evaluated against. Personally, I think that the number of qualified registrations to a user community is a much more interesting way of sizing the user base for a commercial Open Source product, and serving as upstream feed into the sales process. Clarifying the terminology used to describe one’s community would benefit all involved.

How could BPM enable fine-grained entitlement?
This is an issue very dear to my heart. Four years ago, I introduced the idea of using the BPMS as ‘Process Firewall’, as illutrated on this partner website, which transcribes sections of our website circa 2002. The idea for a process firewall is pretty simple: model your process in BPMN using multiple swimlanes, one for each participant, be it a human being, an external system, or another process. The very action of breaking your process down into participants implicitely defines who can do what, which is also called authorization, or entitlement. Essentially, entitlement definition becomes a simple by-product of process design, and it’s one that comes out for free, with the right level of granularity. Back then, very few people understood the idea, but today it seems to be gaining a lot more acceptance. Timing is everything…

How many customers does a typical BPM vendor have?
Most pure-play BPM vendors will claim to have several hundred customers. Some traditional workflow vendors have thousands of customers. But if you’re looking for customers that have used BPM products themselvess without relying on the vendor’s own consulting services, I doubt that very many BPM vendors really have more than a handful of these, if any at all… And you’ll be hard-pressed to find these customers talking freely about their experience on the Internet. This is what we are trying to change. Customers, speak up, please!

When we launched the program, some experienced industry insiders advised us that it might be more difficult than we thought to actually make it work. One of them was Larry Augustin, certainly one of the most knowledgeable experts in the field of commercial Open Source. Larry used to sit on the Board of Directors for JBoss, and shared with me the fact that JBoss had tried to develop such a model in the past, without much success. The fact that others tried it before does not necessarily mean that it’s impossible, but it’s something that one should pay attention to. I did, and gave ourselves about six months to prove — or disprove — the overall model.

What we found is that even though customers intuitively understood the benefits they could get from the model — I won’t describe it in details here, for it’s already explained there, and would be quick to suggest very many projects to us, they had a hard time pulling the trigger and committing hard cash for them. There was some kind of psychological barrier that prevented them to move forward in a timely manner.

When asking candidate sponsors what they needed to commit, the answers we got more often than not was for us to get another sponsor for the same project first. Essentially, they did not want to be first in putting up with the money. This made us realize that one of the rules we set for the program might have created some unnecessary barrier to adoption. Indeed, we decided that in order to prevent the development of customer-specific features, at least two sponsors would have to participate in the funding of any development. While the original idea was legitimate, it put everybody in some kind of catch 22 situation that locked everything up. Once we decided to remove this constraint, at least temporarily, things started to move again.

We cannot disclose the identity of our first sponsor, for we are under NDA with them. That being said, we’re free to indicate that the project is related to the integration of Intalio|BPMS with the Apache ServiceMix ESB. This particular feature had been on our roadmap before the customer requested it, which made the decision to develop it an easy one from a pure product management standpoint, and made the requirement to secure at least two sponsors for it one that we could live without. As this project is being developed, our customer is building a higher level of comfort with the partnership model, and we expect them to sponsor additional features down the road.

Another thing we learned is that the simple fact of sharing our roadmap and gathering feedback from would-be sponsors on our community website helped us assess internally where the actual needs of our users could be, and gave us a very clear direction from a product development standpoint. In that respect, this participative program is acting as a very powerful helper to the product management process, far beyond our original expectations.

At this point, it’s too early to tell whether the program is actually working or not, for we only have anecdotal evidence. That being said, having made it work for one customer proves that it can be done, and that’s all we needed to convince ourselves that we should keep developing it. We’ll keep you posted on the tweaks we’re making to the model in order to make it even more effective.

The idea is pretty simple: generate leads through word-of-mouth and active blogging (happy customers are the best salespersons you can get), consider everybody as a customer (including software vendors and system integrators), sell something affordable initially (training is the key to the whole process), then offer more advanced services to the customers who can most likely pay for it.

Shown on the sales process are expected conversion rates from one stage of the process to another. For example, we expect 1.5% of end users installing our product to sign for a training class within four months, and 25% of those to sign for a support contract no longer than six months after they attended the training class initially.

Also, we make sure that every step in the sales process pays for itself, so that we increase our gross margins and reduce any pre-sales investments to the bare minimum. Sales for our training classes are done through our website, and customers pay by credit card before they receive training. Similarly, sales for support contracts are done over the phone, and we never travel to meet a customer before a support contract has been signed. Following these very simple rules helps us ensure that we develop a profitable and scalable sales process.

We implemented this invested sales process less than six months ago, so it’s too early to tell whether things will pan out as we expected or not, and we certainly will have to make some adjustements along the way. Nevertheless, we started the year with 12 customers, and we now have 70, listed there for the ones we’re allowed to talk about. In other words, we’re off to a good start!

This trend was perfectly explained by Salesforce.com’s CEO Marc Benioff in his recent internal memo published on ZDNet.com by Richard MacManus. What Marc calls the Business Web is essentially a set of services that will provide the tools and infrastructure required for conducting business online. Of course, Office 2.0 is part of this picture, and Marc was kind enough to link to my Office 2.0 setup in his memo. Thank you Marc, much appreciated!

What I find especially interesting in the idea of services replacing software is that you can see this transition at play across the entire IT industry today, not just for Software as a Service providers. For example, Open Source business models developed by companies such as RedHat, JBoss, or Intalio essentially replace software licenses by services, be they for training or support & maintenance.

But make no mistake, a lot of money can be made selling support & maintenance contracts once you tie them to automated software updates. When a RedHat customer buys a maintenance & support contract for RedHat Enterprise Linux, what she really buys is the assurance that the very latest security patches will be part of the next software update. This business is very similar to the insurance business, with one major difference difference: when subscribing to a support & maintenance contract for Open Source software, customers get recognizable value at each software update. Instead, when buying a policy from an insurance broker, customers acknowledge value only when an accident happens. For this reason, the business of selling Open Source software has a lot more potential than the business of selling insurance policies. I like that!

Similarly, when companies buy one of these network appliances for spam filtering or XML acceleration, they neither buy a piece of hardware nor a bunch of software, but rather pay for the convenience of having the right piece of software installed on the right piece of hardware. Essentially, they buy installation and packaging services.

As far as I can tell, most software companies will morph their business into one of the three business models outlined above: Appliances, Open Source, or Software as a Service. The larger traditional software vendors might be able to resist the transition longer than their smaller competitors, but down the road, everybody will have to convert to one of these three models, or to some hybrid models that will have been devised by then.

In this Rock-Paper-Scissors game, service is the winner and the winner takes all.

In a nutshell, we have inverted the traditional sales process for enterprise software. Instead of asking customers to pay for a perpetual license upfront, then selling them support services complemented by some training services and occasional consulting, we make our software available for free. You can download a complete BPMS including a BPMN process modeler, a BPEL process server, a BPEL4People workflow suite, and a BAM framework, entirely free of charge. And not only can you use it for development purposes, you can also deploy it into production, without having to pay us a dime.

Once you played with the software a little bit and got some coaching from our community members, you might decide that you would benefit from some formal training, especially after having read some of the feedback we got from customers who attended one of our regular classes.

After participating to such a training, customers usually do one of two things: either they send more people to training, or they subscribe to a support and maintenance plan. Today, this gives them access to our support engineers on a 24/7 basis. Down the road, they’ll also benefit from automated updates to our product, significantly simplifying the maintenance of complex BPM deployments.

Finally, our most demanding customers will want to upgrade to advanced versions of our product that support other databases than MySQL and provide load-balancing and failover capabilities. In such a case, they will be able to upgrade to Intalio|BPMS Enterprise Edition, which is licensed through yearly subscriptions in order to reduce upfront costs.

Along the way, we sell consulting services to Independent Software Vendors that want to embed our technology within their own products, while partnering with System Integrators for providing implementation services.

For customers, the benefits of such as model are multiple: lower upfront investment, reduced risk of adoption, and scalable deployment model. For us, the main advantage is in reducing barrier to adoption, thereby increasing market share without having to deploy significant sales and marketing resources.

Of course, this sales process inversion leads to increased deferred revenues, which mandates very strong financial discipline. As a result, customers should not be surprised that we allocate our limited resources in a very careful manner, and that they might have to pay consulting fees for spending time with us. But when they understand our model, they immediately realize the benefits they can gain from it, and usually go for it.

Welcome to the brave new world of enterprise software!

First, it takes very significant resources to develop any piece of infrastructure software to be used in a mission-critical environment, be it an operating system (think Linux), a database (think MySQL), a web server (think Apache), or an application server (think JBoss). The Open Source development process has been very effective at producing such software in a reliable fashion. Because the software gets deployed by more users, and because the source code is accessible to more contributors, Open Source offerings tend to outperform any closed source alternative over time. This is true for a process engine as well.

Second, the semantic of process execution at runtime is critical to some applications, especially for compliance requirements or security reasons. When the code of the process engine is open for all to see, no ambiguity can exist as to how processes will actually execute. And if the engine’s developers made an error in interpreting the BPEL specification, anyone is free to contribute a fix that will benefit the entire community of users.

Third, processes are everywhere, yet Workflow and BPM technologies have failed to gain mainstream acceptance, even though they’ve been available for more than 15 years now. One reason for this is that until now, BPM products have been built upon proprietary process execution models. Also, they have proven to be quite difficult and expensive for Independent Software Vendors (ISV) to embed within their own products. The wide adoption of BPEL is fixing the first problem, while the availability of a rock-solid Open Source BPEL engine is expected to address the second.

For all these reasons, we need a good Open Source BPEL engine. Here is one!

It all started with the signing of Intalio’s fiftieth customer. If your company has a hundred, a thousand, or ten thousands of them, it might not mean much to you. But if like us you started the year with only 12 of them, you must know why we think it’s a pretty cool milestone. A full list is available there, courtesy of the excellent Dabble DB, and you can even subscribe to the RSS feed for it if you want to get notified everytime we sign a new one.

Then, I received a nice email from a journalist working for FORTUNE Magazine and asking for an interview in order to better understand how BPM could complement Web 2.0 services such as Basecamp or JotSpot as a way to support collaborative business processes. In and by itself, this was a pretty cool opportunity to help get the message across, but what really made my day was to learn that the introduction had been made by Marc Benioff, CEO of Salesforce.com. If this is not a good example of viral marketing, I do not know what is. The interview went well, and I expect some coverage in a future issue of the print magazine. Many thanks to Marc for this plug!

Finally, the day ended with the second edition of the Open Source wine dinners organized by Andrew Aitken from Olliance and Mark Radcliffe from DLA Piper Rudnick Gray Cary. This one took place at Lavanda in Palo Alto, and gave me the opportunity to meet with the CEOs of other Open Source companies.

Of particular interest was a discussion with John Roberts, CEO of SugarCRM. What I learned is that SugarCRM is not planning to spend much of the $18.77M it raised from NEA last year, which is a great validation of the business model that the company has put together. Much like every SaaS company has a vested interest in Salesforce.com’s success, every Open Source company has a vested interest in SugarCRM’s success. In making our transition back to Open Source, we learned a lot from SugarCRM book, and I wish the company all the success it can get. Note to Marc: don’t you worry! I am sticking to my guns and remain a loyal Salesforce.com customer.

Another interesting topic was a question raised by Sharam Sasson, CEO of Jitterbit, and tackled by Larry Augustin, Chairman of VA Software and board member for many Open Source companies. The question was the following: is the Open Source model for enterprise software only applicable to mature market segments, or could it support the creation of brand new ones?

Larry’s take on it was that new market segments require significant marketing budgets to be spent upfront for them to be created, and this is something that Open Source companies usually cannot afford. Even though the very few examples of successful Open Source companies that are available today — RedHat, JBoss to name a few — seem to validate this assumption, I tend to believe that Open Source is only part of the equation, and a new model for starting enterprise software companies has emerged, which will require much less working capital than has been deemed necessary in the past.

By using new marketing tools such as blogs and webinars, and by lowering the barrier to adoption for customers through the adoption of multiple packaging options such as Open Source software, network appliances, and Software as a Service models, sales and marketing costs can be reduced significantly. Growth will be slower than in the past, but developing a successful and lasting software company has always taken time — just ask Bill Gates or Larry Ellison if in doubt.

Andrew’s and Mark’s dinners always foster fascinating discussions like this one, and I very much look forward to the next one. On to the rest of the week now…

But before we get to the details of the business case, let me clarify a couple of points. First, the aforementioned proposal certainly is a bold one, and should not be considered as an all-or-nothing idea. Instead, it should be viewed as a process that would lead SAP to open source the entire NetWeaver stack incrementally, starting with the most commoditized components, such as the application server and the portal. Second, Open Source is both a development model and a licensing model. Establishing the right developer community and defining the appropriate licensing terms are no trivial efforts. This article provides no guidance regarding these issues, but is an attempt at fostering discussions that will serve as a foundation for future work, which will eventually lead to the open source release of the NetWeaver platform.

There is no doubt in my mind that business applications sold by SAP ten years from now will be built upon an Open Source foundation. The real questions are how long it will take to get there, and how difficult will the transition be for SAP customers. The sooner SAP embarks on this journey, the easier the transition. Here are some reasons why such an Open Source foundation makes sense:

SAP is a process company, not a technology company
SAP executives made it very clear that SAP is a process company. It delivers value to customers in the form of best practice processes that customers can turn into best-in-class processes through the use of advanced tools, which together form what SAP calls the Business Process Platform. In that respect, technology is a means to an end. As Henning Kagermann pointed out, SAP is in the business of solving customer problems. Business applications are developed in support to such solutions, and the underlying infrastructure is developed in support to such applications. In other words, NetWeaver is two degrees of separation remote from the value customers buy from SAP. Something like NetWeaver is required for the solution to be delivered, but it is not where the value comes from. As such, it might as well be free. Not surprisingly, many parts of NetWeaver come at no extra charge with mySAP Business Suite 2005.

Open Source is a great insurance policy for customers
Service Oriented Architecture (SOA), or what SAP calls the Enterprise Service Architecture (ESA), will become the fabric of tomorrow’s IT environments. As such, more and more customers express concerns regarding the prospect of delegating the responsibility of building and maintaining such a critical piece of their IT infrastructure to any single vendor. The lower you move through the IT stack, the least customers are willing to accept any kind of vendor locking, which is one of the reasons why the Linux operating system has been so successful. Above the operating system lies SOA, and this is the next layer of the stack that customers will demand be open sourced. If we believe in this never ending commoditization process, and the role played by Open Source as an enabler and an accelerator for it, the question is no more IF, but WHEN and WHO. Very few vendors will have the scale necessary to support the development of a complete Open Source Service Oriented Architecture. IBM and Oracle will be among them, but SAP could lead the pack.

SAP has little to lose
What makes the idea of an Open Source NetWeaver so attractive is that it should actually be possible for SAP to make such a move without losing much. Unlike IBM and Oracle, SAP derives very little revenue from its infrastructure software. As a result, moving to an Open Source model for it should have limited impact on the balance sheet. Should IBM consider making a similar move, it would alienate both its salesforce and its network of resellers. No such thing would happen with the open sourcing of NetWeaver.

SAP has a lot to gain
The most significant challenge in SAP’s attempt at becoming the leading business process platform vendor is in building an active network of partners that will develop value-added applications on top of SAP’s platform. For an independent software vendor to adopt any platform for building its applications, barriers to adoption and switching costs must be as low as possible, while the market directly addressable through the platform’s adoption must be as large as possible. A successful Open Source strategy is a good way to offer both. If NetWeaver were to be made available under an Open Source license, ISVs that are not necessarily targeting the SAP customer base would nevertheless consider it as an alternative to other Open Source platforms such as Apache Geronimo or JBoss JEMS. As a result, it would extend the range of applications available for the NetWeaver platform, and it would accelerate the development of an ecosystem around it. The return on investment for such a strategy should be higher than any other business development initiative I could think of.

Customers would benefit from it
SAP customers might not pay directly for NetWeaver, but they only get what SAP can develop on its own. While an Open Source development process is by no means a way to reduce engineering costs for a software company — managing an Open Source community requires the dedication of significant resources, it’s a great way to get contributions from others, which over time can make a significant difference. For example, if NetWeaver were to be made avaiable under an acceptable Open Source license, it is likely that collections of connectors to third-party applications would be developed for it by end-users, system integrators and independent software vendors, then integrated back into the platform and made available for free to the rest of the community. Today, customers have to pay for such connectors, usually buying them from vendors others than SAP. The less money they have to spend on such third-party components, the more they can spend on SAP applications.

As more discussions take place around this idea, I expect more justifications for it to emerge. Of course, a lot of objections will be raised too, which will have to be addressed in a pragmatic manner. At the end of the day, SAP is very much a customer-driven organization, and the fastest way for it to embrace an Open Source model for NetWeaver is for SAP customers to explain why it would benefit their business and the business of their own customers. I invite all of them to use this forum as a starting point for such a fascinating endeavor.

I intend to learn more and write quite a bit about SAP NetWeaver, the upcoming Business Process Platform, and SAP’s plan regarding Open Source and Software as a Service. If you happen to be there as well, please feel free to drop me a line and let’s meet for a drink, a motorcycle ride, or a short trip on an airplane. Many thanks to Jeff for the invitation, and kudos to the folks at SAP who seem to really get this blogging thing.

And another Office 2.0 bug got fixed today: welcome ajaxTunes!

Tim Clark suggested ‘Mobility’ as 11th flattener, Jeff Zwelling made a very convincing case for ‘Instant Feedback Loops’ as a comment to my original post, and all participants agreed that the flattening of the world is a process, not an outcome. Sanjay Kalra and Jeff Nolan also had good things to say about the event. I would like to thank everybody who partipated. Also, many thanks to Lohika for their sponsorship of the event.

The next breakfast will take place in the Bay Area sometime in May and will be dedicated to BPM 2.0. If you’re a BPM vendor or a BPM practitioner and would like to contribute to the definition of this new market category, drop me a line and I will add you to the guest list.

Weblogs are great networking tools, but nothing can beat good old-fashioned face-to-face meetings, especially when they come with nice food and drinks. With that in mind, I have decided to organize regular events during which readers of this blog will be able to meet and discuss about recently-covered topics. The first event of this type will be a breakfast to take place at the Hotel Sofitel in Redwood City, CA on Tuesday, April 18, 2006, from 8:00 AM to 10 AM.

The theme for this meeting is The World is Flat, based on Thomas L. Friedman’s excellent book. We will discuss how Friedman’s ten flatteners can be put to use by software vendors today, using Intalio and a couple other start-ups as reference cases. This event is sponsored by Lohika — our offshore development partner — and will feature their CTO, Mark Phillips, who will share his experience working with small and large software companies that successfully offshored part of their software engineering process to Western Ukraine.

If you work for a software company and want to learn about hybrid business models that blend open source and commercial software, understand the benefits of offshore development beyond simple cost reduction, become more familiar with our Demand Driven Development process, get a preview demo of the upcoming Intalio|BPMS 4.1, or simply eat the best croissants you can get in the Bay Area, please register here and join us to this free breakfast.

And because I would fail my duty as Chief Office 2.0 Evangelist if I did not inject some neat online service into the mix, our registration form was developed with the ultra cool Wufoo form editor. I might very well be the first production user for this service, which still is under heavy development as I am writing these lines, and the good folks at Infinity Box literally worked around the clock from Sunday night to Wednesday morning to give me access to a preview release. I cannot say how much I appreciate the effort, and am eagerly waiting for the public release. In the meantime, all I can tell you is that Wufoo rocks and that I have added a new project to my Office 2.0 Mashup List that consists in integrating Wufoo with Salesforce.com. If that sounds like fun to you, come give us a hand in developing one of these mahsups.