Anyone who wasn’t able to attend the InfoWorld Virtualisation Executive Forum in New York recently missed out on a fascinating event. A panel discussion I moderated on virtualisation and Linux demonstrated that the open source community remains very much interested in and engaged with this topic. But one thing that struck me and several of my colleagues, based on audience reaction to the various sessions, was just how early we still are in the lifecycle of virtualisation technologies.
A case in point is the convergence of server virtualisation and storage virtualisation. Vendors and analysts seem to agree that while these technologies aren’t closely related under the hood, they do support and feed into each other because they serve similar goals.
Nonetheless, customers tend to view server and storage virtualisation very differently. In a lunchtime presentation, ’s vice president of marketing, Paul Calento, shared the results of a recent InfoWorld survey of IT managers. The majority of participants in that survey responded that server virtualisation could be achieved relatively easily and inexpensively, and that they plan to do it using best-of-breed products from a variety of mainstream vendors. However, the same respondents felt that storage virtualisation can only be achieved expensively and with difficulty, and they expect to use a solution from a single vendor.
These results can hardly come as a shock to anyone who has ever managed enterprise storage, but they’re interesting all the same. Clearly, open source plays a role in this. While there are plenty of server virtualisation solutions and tools that support or are based on open source, there has been comparatively little movement on open source storage virtualisation.
Therefore, I took the opportunity at the Virtualisation Executive Forum to sit down with Brian Stevens, the CTO of Red Hat, to learn if anything was being done about this. Like Novell, Red Hat is very involved in bringing Xen virtualisation technology to its enterprise Linux distribution. I wanted to know if anything similar was planned for storage.
Unfortunately, Stevens’ answer was no — at least not yet. While he pointed out that technology like Red Hat’s open source GFS (Global File System) can help virtualised server environments make more effective use of storage, higher-level storage virtualisation software remains beyond the capabilities of open source developers.
That could be changing, however. Creating open source storage software is difficult today because doing so requires access to a broad range of storage technologies, many of which are proprietary and undocumented. Independent developers just don’t have the resources to tackle the problem. But the desire for cost savings is driving the industry towards a new set of technologies and standards — including iSCSI and
10 Gigabit Ethernet — that are considerably more open.
As we’ve seen in other segments of the software market, these open standards are the key. Once they’re in place, developers can begin to build on their foundation. It will take time, certainly, but Stevens doesn’t think storage will stay immune to the open source momentum forever, and neither do I.