Open Source Commentary from Navica's CEO, Bernard Golden

In This Issue

• Oracle's Open Source Acquisition Spree: Is it the Big Bad Wolf?

• Navica News

Oracle's Open Source Acquisition Spree: Is it the Big Bad Wolf?

The hot gossip in the open source world over the past couple of weeks has been Oracle's purchase of Sleepycat and putative purchase of JBoss. Much of the news coverage about these purchases is whether their users – in other words, the community, should fear the implications of a purchase. People are worried that this exemplar of proprietary software might be planning to hurt JBoss and, in a kind of collateral damage, hurt JBoss' user base. In other words, mighty Oracle may be planning to crush JBoss for its own nefarious ends. In a slightly more benign interpretation, Oracle may not be planning to harm JBoss, but may end up smothering it as it is absorbed into the giant corporate clutches of Oracle. No matter what Oracle's motives, the end result is JBoss – and its community – being irreparably damaged by an Oracle purchase.

But is that the most likely outcome? In this newsletter, I'd like to offer a different interpretation of the likely outcome of an Oracle acquisition of JBoss – one that is far more threatening to Oracle and favorable to JBoss and its community. In my view, rather than the dory of JBoss being crushed by the ocean liner of Oracle, a more likely scenario is the Titanic of Oracle being holed – perhaps mortally – by the iceberg of JBoss.

Oracle: A Threat to Open Source?

Many people believe that the open source community is threatened by what Oracle might do with the JBoss after it buys the company. Could it actively seek to retard the technical progress of the product so as to favor its own, market-trailing application server? Oracle's purchase would in this scenario be a classic predatory acquisition in order to eliminate competition.

There is much to recommend this perspective. Despite all its efforts, Oracle's app server has little presence in IT infrastructures. It seems to have a decent market share, but that is mostly achieved by Oracle bundling its app server as part of its other products – you buy the database, and they throw in the app server for free. Oracle utilizes its app server in its database management infrastructure, much to the detriment of ease of use. By basing their management infrastructure on their app server, they have achieved what previously seemed impossible – they've made the database even harder to install.

Therefore, in an attempt to build its own app server business, Oracle would eliminate a significant competitor in the guise of JBoss and gain an opportunity to grow its market share and go after other competitors, most notably BEA.

However, arguing against this view is two things: Oracle's stated intentions, and the fact that JBoss is an open source product.

Regarding Oracle's intentions, Larry Ellison has addressed their motive regarding open source purchases twice. In the first, he noted that Oracle is moving from a license sales business model to a subscription model – in other words, from a growth model to a harvest model. Oracle would purchase JBoss in order to have another product to sell into its installed base. Ellison characterized this new business strategy as an “all you can eat” model.

In a later statement, Ellison noted that Oracle is pursuing open source because it recognizes that open source adoption is growing rapidly – and Oracle doesn't want to try and stand in the way of the trend.

Consistent with this view is another motive imputed to Oracle's purchase – the desire to use JBoss as a “loss leader” available for use by developers, with an upsell to Oracle's own app server when the newly-developed system is ready to go into production. Making the open source JBoss app server available would ultimately drive additional software license revenues.

Should One Believe Oracle about its Open Source Intentions?

However, one may not be convinced by Ellison's statements. They may be a smokescreen designed to camouflage its true intentions. If its real motive is to eliminate a competitor, perhaps JBoss users should be worried. After all, anyone who has watched the aftermath of the Oracle PeopleSoft/Siebel acquisitions can understand this concern.

Companies that purchased PeopleSoft or Siebel, making a strategic commitment to those products, now anxiously await their future. Oracle has announced “Project Fusion” which is to deliver a mashup of PeopleSoft, Siebel, and Oracle functionality. You don't have to be a fortune teller to recognize that, Oracle reassurances to the contrary, Fusion is likely to be an expensive, difficult, and protracted exercise that is bound to cause enormous suffering to its “beneficiaries.” For these users, Oracle's acquisitions have not been a good thing.

However, no matter what Oracle's intentions, in the event of an acquisition, the JBoss community would have a vastly different experience than the PeopleSoft and Siebel user base; indeed, the cause of the difference can be seen in the terms used to describe the two populations: community and user base.

The Difference Between a User Base and a Community

A user base is made up of companies that have purchased a software product with a typical proprietary license – one that restricts their ability to manipulate the product and thereby leaves them in thrall to the vendor. Consequently, former PeopleSoft and Siebel customers can only wait and hope that things will turn out well for them; they have no leverage over the vendor.

By contrast, an open source community springs up and surrounds an open source product, enabling self-reliance of users – all based on source code availability, of course. If the JBoss community doesn't like the direction that the commercial entity is taking the JBoss product, it can pick up its marbles and leave – forking the source code and starting anew – leaving the commercial entity all alone with no one to play with. By its very nature, open source software provides its users with leverage against vendor intentions.

Consequently, Jboss users have far less to worry about regarding Oracle's intentions for their product than their PeopleSoft/Siebel counterparts. Their product will continue in a positive direction – and if it doesn't, they can redirect it.

Actually, the Threat Shoe is on the Other Foot

Fine, Jboss users would be protected in the event of an acquisition, but what about Oracle? It takes two to tango, and perhaps Oracle wouldn't actually be leading this dance.

There have been a couple of open source acquisitions by proprietary vendors to this point, and what is especially striking about them is how different they are from typical software acquisitions in the past. When proprietary vendors purchase other established proprietary vendors, the acquisition price is usually set at some multiple of revenues, typically two or three times.

In the case of the open source acquisitions, however, the prices have been astronomical in light of previous acquisition benchmarks. The rumored price for Jboss was $400 million; this for a company with revenues surely no higher than $30 or $40 million dollars, and possibly much lower – clearly way out of bounds, given historical norms.

In other words, these deals make no economic sense in the context of a proprietary software model. Larry Ellison made mention of this in a discussion about a previous, unconsummated, purchase attempt of MySQL. He noted that MySQL's $40 million in revenues would be minuscule compared to Oracle's overall $15 billion yearly run rate.

Open Source Purchases: A New Yardstick for a New Software Era

Therein lies Oracle's problem. They are applying historical yardsticks to tomorrow's business models and can't see how to make the measurement come out right. Of course, this didn't seem to dissuade them from purchasing Sleepycat, for which the numbers were, presumably, not significantly better. Perhaps Oracle feels that, if they purchase Jboss, they can achieve maintenance revenues on the products similar to the fees they charge for other Oracle products. If so, they have a lesson coming in price elasticity of demand. Jboss maintenance and support is significantly less expensive than Oracle's – and Jboss users are unlikely to accept a higher price just to bring Jboss maintenance prices into line with Oracle's pricebook.

The acquisitions that have occurred thus far must have been justified on strategic, rather than economic grounds. The boardroom discussions must have gone something along the line of “this is the coming wave of software – we should make sure we have exposure to it to ensure we address our customer needs.”

Can an Open Source Purchase Succeed within a Proprietary Company?

History has not been not kind to companies that attempt to offer two different business models within one entity. Clayton Christensen has addressed this situation extensively in his books on innovation. One example he offers that has striking parallels with the potential Oracle/JBoss marriage was the impact of discount stores on variety stores and how two different companies responded.

Without recounting the entire story (Christensen describes it in in pages 110 – 115 of The Innovator's Dilemma), he notes that two variety store chains, Kresge and Woolworth, both decided to pursue a discount chain strategy. Kresge set up a entire different division, bringing us the blue-light specials of Kmart, while Woolworth chose to launch its Woolco initiative within its mainline Woolworth division. Kmart prospered, while Woolco eventually was shut down.

The interesting thing is why Woolco failed. According to Christensen, it wasn't because the formula of retailing wasn't understood by Woolworth's, or because of a shortage of capital, or any other clear handicap. It was because Woolworth's imposed its business expectations, especially regarding pricing and gross margins, upon its internal Woolco operation. Essentially, it converted Woolco from a discount operation to a less-impressively executed version of Woolworth's, with predictable results.

The longer term outcome was also predictable. The very business model Woolworth's was imposing upon Woolco was eventually crushed by the discount model it was unable to execute successfully. Woolworth's hung on for a number of years, but eventually was transformed into the Foot Locker chain.

What does this mean for Oracle?

In the event of a Jboss acquisition, Oracle will attempt to offer a mixed menu of proprietary and open source products. This will not be successful (and, as noted above, attempting to sell Jboss products at Oracle prices would fail as well). Eventually, Oracle would be forced to discontinue the Jboss initiative, or perhaps, spin it back out. The corporate body would reject the open source transplant.

While never rising to the level of overt internal resistance, Oracle personnel would ignore the Jboss product – it would get poor promotion, the sales force would gloss over it, the least talented and influential marketing personnel would be assigned to product manage it. Simply put, Oracle would impose proprietary expectations on Jboss and thereby fail at being an open source company.

I have had personal experience attempting to shepherd lower-margin products in a high-margin company. While working at Informix, I was assigned responsibility for developing the database on the just-released Windows NT platform. At release time, I was unable to generate any enthusiasm for the product within the rest of the company.

In an attempt to get corporate support for the product, I met with Informix's then-CEO, who observed that Informix was a complex product that required a lot of technical support, and the prices necessary to sell the database on NT wouldn't provide sufficient margin and therefore the sales force wouldn't get behind it. In fact, he said, NT probably wasn't going to make much of a mark in IT organizations. Or, put another way, he didn't want to make the business practice changes necessary to make a success of the product.

The rest of the story is obvious. The product languished – it remained on the pricelist and got a bullet point on the sales presentations, but never went anywhere.

Of course, as discussed earlier in this piece, all of the internal Oracle back-and-forth that would accompany a Jboss acquisition would be irrelevant to Jboss users. They would be protected by the open source license Jboss carries and the ability to fork the code base.

However, the true impact of a potential Jboss acquisition can be seen in the rest of the Informix/Windows NT story. Contrary to the CEO's prediction, NT went on to be a huge portion of IT infrastructures. Informix, on the other hand, eventually was sold at around one times revenues to IBM. The lesson is clear: you can resist business model shifts, but they are far stronger than you.

Far from being worried about what Oracle will do to Jboss, the real threat is what Jboss (and its open source brethren) will do to Oracle (and its proprietary brethren). To offer another historical allegory, the proprietary vendors are lumbering Titanics that will inevitably founder on the icebergs of open source.

Navica News

You can hear me speak at these upcoming events:

March 17, 8:30 a.m.: SDWest, Santa Clara Convention Center. Presentation topic: "Open Source ROI: Achieving the Promise of Open Source." See www.sdexpo.com for more information.

April 27, 7:00 p.m.: Peninsula Linux Users' Group. Presentation topic: "GPL3 -- Where Free Software is Going." See www.penlug.org for more information.

Also: Look for a new GPL3 blog posting on www.cio.com in the next week or so. Look for it under Tech LinkLetter in the blogs section. Topic will be "Richard Stallman: Entrepreneur of Freedom."