Monday, February 19, 2007

What is a startup to do?

Another post motivated by happenings in the Virtualization space! I met some RedHat folks during the RSA conference a few weeks ago and quizzed them about the new Kernel Virtual Machine (KVM) that Linus merged into the main Linux Kernel. The answer was surprisingly candid and straightforward and was reinforced by a recent announcement by RedHat's CTO. Expectedly, RedHat is hedging its bets by bundling both Xen and KVM in its Fedora Core line of products.

Apart from the fact that RedHat seems to have good internal communication, this announcement has significant ramifications for all the big Virtualization players. Xen worked really hard to get RedHat to package it by default, hoping that it will help them establish a big footprint in the Linux market. They even had some disastrous PR failures when RedHat claimed that Xen was not ready for primetime (see here). And finally, just when their effort was nearing fruition, Qumranet sprang a surprise by getting its KVM technology endorsed by Linus and RedHat.

Though Xen has first-mover advantage as the default virtualization technology available in RedHat and is probably more mature than KVM, this news could not have come at a worse time. Bigger and more mature (read risk-averse) IT organizations that were looking at Xen on Linux will now prefer waiting for KVM to be available before investing heavily. Smaller enterprises may also pilot KVM to see if it will satisfy their needs or at the very least slow their adoption of Xen.

Now, what should Xen do faced with such a situation? They may still have some good ways of fighting VMware/Microsoft from upstream and KVM downstream -
  1. Learn from VMware - VMware was faced by a similar assault in 2005-06 from Microsoft and Xen. Xen was downstream and offering a free product and Microsoft, though not technically upstream, used its vantage position to make deep price cuts. VMware responded by giving away significant parts of its base virtualization platform for free and building out Management Infrastructure software that it is now the primary cash cow.
  2. Drive adoption of the platform in applications - One of the often underestimated drivers of VMware's growth its integration with applications. VMware has successfully evangelized many vendors (Vizioncore, Surgient, NetExam, Opsware, Leostream, etc) to use its APIs to build their application. Though this is arguably a small channel for VMware, Xen should not ignore this altogether as this market has a lot of potential. VMware has had better success in this area than the open-source Xen - can they do something about this?
  3. Don't lose track of your endgoal - winning a slice of the Virtualization pie. Face it, Xen - you came close to winning the Linux market, but looks like you will have to suffer a huge heartbreak! You'll never own the low-margin Linux server virtualization market - probably the best thing to happen to you. Stop running behind Linux and instead focus on out-innovating VMware in one or two niche areas on Windows and Solaris.
  4. Acquire smaller players - Xen should look at acquiring Virtual Iron or other smaller players to help fasttrack your move into Management Infrastructure or Application support.
Do you have any other ideas? Send them to me

No comments: