Is an Open Source Approach to Virtualization Management in VMware’s Future?

One of the most interesting aspects of analyzing a vendor’s product strategy is to look at how a vendor uses an existing position of strength to embark upon initiatives that attack and weaken a competitors existing position of strength. These attacks upon a competitor do not usually come in the form of a direct attack upon a revenue generating product with a similar “better” product, but rather most often take the form of attacking the very business model and revenue stream that sustains the current and future development, marketing and sales efforts for that product. In other words these attacks are designed to deprive the competing vendor of the oxygen (revenue) that is needed to sustain and improve the product in the marketplace. Some excellent recent examples of such attacks are:

  1. As Netscape became successful, Microsoft responded by making the browser a free feature of the operating system. This was at the time just the latest example of Microsoft “improving” either Windows or Office by taking something that had been a standalone product and making it a feature of one of its flagship offerings. If you are old enough you might remember a time when spell checkers were sold independently of word processors – and were then bundled into word processors for free. Microsoft has pursued initiatives like this countless times to its advantage. TCP/IP stacks used to be products, now they are features of operating systems. Directory Services used to be a product, now they are features of operating systems.
  2. As VMware became successful, Microsoft tried the same tack described above with virtualization. It essentially bundled virtualization into the Windows Server operating system and gave it away for free. Only this time it did not work (at least not yet). The reasons why will be the subject of other posts that will delve into this subject in detail.
  3. Google is working hard to deprive the Windows and Office businesses of oxygen by offering Google Apps, and Google AppEngine on terms that are extremely cost effective compared to licensing Windows and Office from Microsoft. Of course this is only possible because Google as a massive revenue stream from advertising to tap for these initiatives.
  4. Since VMware is run largely by ex-Microsoft executives, it should not come as a surprise that VMware would embark upon some oxygen deprivation tactics of its own. The first of these was VMware’s efforts to offer an applications runtime environment through the acquisition of SpringSource. This was a hugely significant step as it put VMware in the position of offering a Java run time environment glued into vSphere. This is a direct threat to Microsoft Windows and Microsoft .Net as someone who chooses to run a Java based application on the Spring TC Server in a VMware VM does not need to buy any systems software from Microsoft. Nor do they have to buy any software from Red Hat, Oracle (who now owns BEA Weblogic) and IBM (the vendor of WebSphere). VMware has subsequently made a series of open source acquisitions designed to further bolster the richness of their open source application server environment. These include RabibitMQ (open source messaging middleware for Java based applications), GemStone (in memory caching of database data), and Redis which helps the performance of multi-user database applications. Taken together this is an oxygen deprivation initiative targeted at Microsoft, Red Hat, IBM, Oracle and anyone else who has either an operating system or an applications run time environment.
  5. The second open source oxygen deprivation initiative from VMware was just announced last week and is covered in our post, vSphere 4 – now with free SUSE Linux.  This is again a direct attack on Microsoft and Red Hat. Now a Linux OS is free with vSphere. So there is absolutely no need to go get a Red Hat support agreement for the Linux OS that underlies the Spring TC server in a guest VM.
  6. Along with the SpringSource acquisition came Hyperic. Hyperic was a successful open source availability and performance management solution targeted at very large scale deployments of web servers (think a hosting provider with tens of thousands of web servers). It is unclear at this point what VMware intends to do with Hyperic, but the best guess is that it will play a role in making applications built on the Spring platform more manageable. A good piece of supporting evidence on this front is this blog by the founder and CEO of Hyperic, Javier Soltero.

However, there is a more important ramification to all of this than just how VMware is competing with Microsoft, Red Hat and others in the systems software (operating systems and applications platform) businesses. That ramification is hinted at Javier Soltero’s comment about DevOps in his post, and in the open source approach to DevOps. While open source is a way of developing and delivering software to customers, DevOps is concept – a concept that has the potential to fundamentally disrupt the management software business. This is the thesis of this post. That VMware just might be on a path to implement DevOps in a set of open source products. Before we look into the ramifications of this, lets explore what DevOps is all about.

It is the case for the vast majority of enterprises, that there is a wall between the people who care about applications, and the people who care about the infrastructure for the applications. The infrastructure people are IT Operations and they care about operating the infrastructure in a cost effective and efficient manner. The applications people are typically the teams that develop applications and that support them in production. This has proven to be an enormously expensive wall. Our recent post, Root Cause Analysis – The Key to Virtualizing Tier 1 Applications, is really about how an enterprise can get a unified view of the applications and the infrastructure and trace problems in applications performance to either the problem in the application or the problem in the infrastructure. The short answer is that despite the fact that most enterprises spend a fortune on hundreds of management tools; this is not easy to do at all. DevOps is all about destroying the wall between the Development and the Operations teams so that they work together on the same problems with the same tools in real time. DevOps is joined at the hip with Agile Development as Agile Development processes put new code into production so quickly and frequently that the wall between Development and Operations must be torn down in order for these applications to work well for their users.

VMware has made its intentions regarding being a management software vendor known on numerous occasions. VMware believes that virtualization and the cloud are catalysts that create an opportunity for a new management stack – and VMware intends to be the vendor of that stack (at least on its own platform). To this end, VMware has already offered products like LifeCycle Manager, Lab Manager, CapacityIQ, and AppSpeed. VMware also acquired four products from the Ionix division of EMC. Therefore VMware seems to be building a conventional set like the rest of the management products in the world, only one that is at its core virtualization and cloud aware.

But what if VMware (potentially in parallel with its product efforts) is up to something fundamentally more ambitious and disruptive? What if it intends to use open source software as a disruptive economic force, and the entirely disruptive concept of DevOps (at least on its own SpringSource TC Server Java platform) to fundamentally disrupt the management software industry? If we look at VMware’s actions towards competing OS vendors the strategy can be neatly summed up as “Those products are no longer necessary”. What if VMware is going down a path to not only compete with the existing management vendors (CA, IBM/Tivoli, HP, BMC and hundreds of others), but to completely invalidate the very premise upon which their products are built and therefore the need for their very existence? Both Rod Johnson and Javier Soltero have been previously quoted and focusing upon automating and improving the “Build, Run, Manage” process which is exactly what DevOps is focused upon as well. Another interesting clue comes from the announcement by Opscode today of the commercial availability of their open source, SaaS delivered mass configuration platform – Chef. Javier Soltero is again quoted in this article saying ““Chef is an important tool for automating infrastructure management and we look forward to its continued growth and success.”


VMware has already demonstrated a penchant for using open source technologies to fundamentally disrupt the value propositions for the products from competing vendors in the systems software and applications platform businesses. This has put the operating systems businesses at Microsoft and Red Hat, and the applications platform businesses at Microsoft, Red Hat, IBM and Oracle under pressure, by providing a cost effective and fully functional alternative to the traditional licensed software models of these companies. It is entirely possible that VMware will pursue the same approach in the management software industry thereby disrupting the business models and product positions of CA, IBM/Tivoli, HP, BMC as well as many of the vendors currently in the VMware ecosystem. Of course the management software vendors are not sitting still. Many of the VMware focused vendors have already broadened their offerings beyond VMware vSphere itself, and the big 4 certainly have an enterprise footprint that will not be easily displaced.