Jeff Clavier's Software Only

Two great blogs to track the enterprise software market

Even though I no longer focus on enterprise architecture and related software solutions, I try to keep an eye on developments in that market, especially as consumer applications and Web 2.0 concepts make their way into it (dubbed Enterprise 2.0 of course)

I have recently added to my “Favorites” folder the 451 CAOS Theory, a group blog written by analysts of the 451 Group, an analyst firm, and Confused of Calcutta, the blog of JP Rangaswami, the CIO of Dresdner Kleiner Wasserstein.

I have met JP a number of times, mainly during conferences in the US and in Europe, and have always enjoyed his progressist approach to the adoption of new tools and processes to Enterprise IT. His “About this Blog” section says it all:

I believe that it is only a matter of time before enterprise software consists of only four types of application: publishing, search, fulfilment and conversation. I believe that weaknesses and corruptions in our own thinking about digital rights and intellectual property rights will have the effect of slowing down or sometimes even blocking this from happening.

I believe we keep building layers of lock-in that prevent information from flowing freely, and that we have a lot to learn about the right thing to do in this respect. I believe identity and presence and authentication and permissioning are in some ways the new battlegrounds, where the freedom of information flow will be fought for, and bitterly at that.

I believe that we do live in an age of information overload, and that we have to find ways of simplifying our access to the information; of assessing the quality of the information; of having better tools to visualise the information, to enrich and improve it, of passing the information on.

I believe that Moore’s Law and Metcalfe’s Law and Gilder’s Law have created an environment where it is finally possible to demonstrate the value of information technology in simple terms rather than by complex inferences and abstract arguments.

I believe that simplicity and convenience are important, and that we have to learn to respect human time.

I believe we need to discuss these things and find ways of getting them right. And I have a fervent hope that through this blog, I can keep the conversations going and learn from them.

As to the 451 CAOS, which stands for Commercial Adoption of Open Source, I really appreciate some of the insights that the 451 analysts share in extenso on the blog, which has this description:

This blog covers open source in the enterprise. It is written by analysts from The 451 Group, an independent technology-industry analyst company focused on the business of enterprise IT innovation, which also produces the 451 CAOS Research Service.

Belated congratulations to Marc Fleury for the $420M sale of JBoss

Marc Fleury has been famous since the start of JBoss for being non-traditional and aggressive - leading his company to a 35% market share in the application server market, outperforming larger companies such as IBM and BEA. After rumors of a potential Oracle acquisition, last week saw the Red Hat acquisition of JBoss for an outstanding $420M consideration ($350M in cash and stock, and a $70M earnout). Matrix Partners, Accel Partners and Intel Capital were the three backers of the company, having invested $10M in the company back in 2004 - raising eyebrows from the community at that time - remember JBoss is “free” (as in open source free).

From a multiple perspective, the acquisition price is an impressive 25X of last year's revenue, and 8x of this year's forecast. Not bad in a world where traditional enterprise software companies sell for 4 to 5X revenues. As David Berlind puts it in this piece, this is a validation of the open source business model for software companies - that manage to reach a leadership position in their market.

Congratulations to Marc (who shares his thoughts on the acquisition here) and his wife and business partner Nathalie, and investors David Skok (his views on JBoss' success here) and Peter Fenton.

Tags: jboss

Don't be an Open Source Ostrich

Lots of companies are currently analyzing the impact of a decision to move from closed source to open source (like Intalio just did). And it is critical to consider the legal implications of these decisions, especially when it comes to the type of license to use, what software your company already leverages, etc. You want to avoid the “whoopsy” moments that happen when you discover that you have been shipping some modified GPL code without contributing back, etc.

I found this essay from lawyer Dennis Kennedy coming very handy in that context. Do take a look, it is clear and well written, and provides useful background. Here is a summary of his checklist:

1. Understand the Different Approaches That the Open Source Licenses Take.
2. Pay Special Attention to the General Public License.
3. Remember the Source Code.
4. Make Reasonable Comparison with Commercial Software.
5. Think in Terms of Choosing, Rather Than Negotiating, Open Source Licenses.
6. Do Not Confuse Open Source with Public Domain.
7. Inventory and Assess What You May Already Be Using.
8. Open Source Use Requires Open Source Training.
9. Reasonable Policies and Procedures Are Not Optional.
10. Treat Open Source Policy as a Team Game.

My favorite quote: “Don't be an Open Source ostrich”, which refers to the sort of default reaction of executives when you bring up the notion of open sourcing portions of the stack a company has spent years and hard cash to develop. As Dennis puts it in his conclusion:

Open Source software is not likely to go away nor are you likely to avoid it. As always, “be prepared” is the best motto. Addressing this area from a reasonable knowledge base, with your eyes wide open, only makes good sense in today's business environment. These ten tips will help you get your Open Source house in order and pave the way for effective and wise use of Open Source software with your legal risks kept within your level of comfort.

Thanks to Alex Muse for the pointer.

Blogging the BPM market

Before moving to the wonderful world of the consumer Internet, I have built, for about fifteen years, mission critical systems in the financial services market - in which you spend a ton of time (and money) integrating “stuff”: front-office to back-office, buy-side and sell-side, transaction processing to risk management, etc. Originally using FTP or manually-written socket-based message transfers, these integration projects gradually started using Enterprise Application Integration (EAI) frameworks such as Tibco's ActiveEnterprise. These frameworks generally started by offering a messaging layer (a Message Oriented Middleware), on top of which were built gobs of features such as message transformation, rules and workflow management, portals, distribution and supervision frameworks, etc. Back in 1999/2000, a new element was added to the stack: Business Process Management, that was supposed to bring together business analysts and process experts on one side, and developers on the other: analysts would draw process diagrams that would automagically be turned into executable processes that developers could deploy within the enterprise, and across enterprise boundaries.

The concept was interesting, and a number of startups got funded in the nascent market (Lombardi, Fuego, Intalio, Savvion,...) that proved to require more time (and more money) than expected to mature. I became friends with Ismael Ghalimi, the co-Founder of Intalio, and made a small personal investment in their Series B back in 2001. Whilst I am not too sure of the “marked to market” value of that investment, Ismael's insights on the BPM and software stacks at large have been most valuable to me over the years. It is therefore good news that he has decided to start a blog called IT|Redux that covers Business Process Management and a lot of surrounding technology. The latest post describes the “open sourcing” of the Intalio framework, an interesting read:

Intalio has now transformed itself into “The Open Source BPMS Company.” Step One was the acquisition of FiveSight Technologies, the company that brought to market the first open source implementation of the BPEL 2.0 specification. Embedded by major open source projects such as the ServiceMix Enterprise Service Bus (ESB) and leading development tools like Sun's Java Studio Enterprise, the FiveSight PXE BPEL 2.0 engine – rebranded Intalio|BPMS – is effectively the most widely deployed BPEL product today. Intalio is planning to continue the development of the BPEL engine under the open source Common Public License, and release its BPMN process modeler and BPEL4People workflow component under similar open source licenses later in 2006. The Intalio|BPMS Open Source Edition includes these three components plus open source ESB, integration adapters, and rule engine. Its customers will be mainly independent software vendors, who will embed the engine and tools into their own end user offerings.

If you have never heard of EAI, no worries: the more modern lingo is ESB (Enterprise Service Bus), SOA (Service Oriented Architecture), etc.

Tags: BPM

Open source and the end of traditional sales and marketing

Larry Augustin (founder of VA Linux, creator of SourceForge.net, and board members of a handful of open source leaders) makes a number of good points in his piece "Time for a New Software Model". This pro-open source pitch is not new per se since Larry has been talking about the end of traditional sales and marketing in that context.

A few point of interest, even though I recommend reading the whole post:

• The problem is that the traditional enterprise software business model is broken. A rabid search for new customers and revenue growth has caused sales and marketing costs to spiral out of control. In fact, Rick Sherlund at Goldman Sachs estimates that in 2005 software companies will spend 82 percent of new license revenue on marketing and sales efforts. That's up from 66 percent in 2000.
• The Open Source model turns the marketing problem on its head. Customers can look at, evaluate and review software without contacting the company that will sell it to them. [...] The customer says, "I want something like that." He locates the Open Source version of the product, downloads it and is using it before the company is involved. [...] the Open Source solution, the CIO is happy with the software. But after using it for a while, begins to wish for documentation, a live-person to ask questions, a phone number for support, and so on. At that point, the customer calls the company saying, "I've been using your product for a year and now I need your help."
• The Open Source model likely delivers at best a 50% cost advantage in R&D. Most Open Source companies gain little community leverage developing the core of their application. R&D community leverage tends to come from testing, bug fixes, and interface/integration code. On the other hand, the savings in sales & marketing could be closer to 75 percent. Without the need to pay for large up-front sales & marketing costs, the vendor doesn't have to charge a new license fee up front. No more charging the customer to sell to them - the customers have to sell themselves - but they still come to the vendor for maintenance and support. And maintenance and support is how most software vendors make their living these days anyway.
• New license growth in enterprise software companies today is stagnating. By changing the revenue mix from primarily new licenses to primarily maintenance on already proven products, a typical enterprise software vendor's top line would drop by 25 to 30 percent. But the decrease in price and increase in market size will enable growth in those previously unreachable markets, creating a vibrant and growing company that will eventual pass its former revenue numbers.

The success of a number of companies like JBoss, MySQL and the potential of SugarCRM and Asterisk show that the model has some legs, and can build sustain companies with 40 to $50M in revenues. It will be interesting to see whether they can grow by an order of magnitude on the same model.

OpenSolaris: Sun's move is gutsy, but is it too late ?

David Berlind over at ZDNet has published an excellent note regarding Sun "open sourcing" Solaris 10: Will Sun's 1600 patents suck the life out of Linux?. It is a thoughful complement to this CNET piece, that goes into more details on the background of the decision.

Sun has been struggling since the end of the bubble with an ever declining market share (and mindshare), and many had "buried" the company (even if it still has over $3B in cash on the balance sheet) because of its reliance on a proprietary version of Unix, and a proprietary chip design. The lack of support for cheap Intel boxes has always been a major issue for ISVs that, in an ideal world, would have wanted to run their Solaris applications on low cost servers. I am discounting the x86 version of Solaris that Sun released in the mid 90's because it was too expensive, slow and lacked a proper ecosystem (tools, drivers, etc.).

I should mention that I have been a big fan of Sun hardware and OS since the late 80's when the startup I had the priviledge to be involved with, Effix Systems, led the adoption of Sun workstations in trading floors in Europe, in the late 80's. We were amongst the first to port our applications to Solaris 2.0 when it was still in alpha stage, because we had won a major trading floor on Wall Street, and the CIO of the bank wanted to deploy on the new OS, and not the old, well proven, SunOS. Even though it was ugly, we had a great experience working with Sun labs to get patch and fixes to deliver and deploy (more or less) on schedule. And heck, at that time, even Scott McNealy was calling us to make sure that everything was going smoothly (and then he dropped a bomb on us when, a couple of years later, he abandoned Xview in favor of Motif without any real migration path, but that's another story).

Back to the subject, Solaris has always been a very good O/S, bit heavy - or one could say feature rich, but certainly one that scales well on the high end. So will we get the best of both worlds ? A great O/S, at a great price point, with the open source community behind it ? For one thing, it could help solving the perceived issue that Solaris is not as good on PC architectures as on Sun boxes. Also, CIOs might be quite keen to understand how real Sun's indemnification and them throwing in their 1600 patents will be. But if the fine print in the license does not introduce too many rinkles and limitations (see GrokLaw's thread for more on that very topic), it might well be a turning point for Sun if (big if) they become a viable and respected alternative to a Red Hat or Suse distro in the LAMP world (SAMP anyone ?).

Berlind's last paragraph summarizes the situation quite well:

What this means in the bigger picture is that all of those developers who are risking patent or copyright infringement by participating in largely unindemnified open source operating system projects such as Linux can pursue their passions in the OpenSolaris community without the threat of a legal sledgehammer swinging randomly around their heads. Whether or not OpenSolaris will be like a big vacuum that sucks the developers right out of the Linux community remains to be seen. But, if I were a Linux developer, right about now the grass would be looking awfully green on the other side of the fence. Is it too good to be true?

Private message: Duncan, do you remember that Sun briefing in 2001, where we suggested that they should open source Solaris ? Was not such a stupid idea after all.