The big question in virtualisation today is whether Microsoft can seriously dent VMware's lead in the virtualisation market. Industry observers are starting to argue that Redmond has that capacity and that it's starting to become evident - but are they right?
The virtualisation business - by which I mean the one that VMware revived in 1999, not the mainframe technology that IBM invented in the 1960s - has in some ways been a fairly staid one since its inception. VMware has dominated the market, even if the penetration rates of virtualisation technology remain low - common consensus is that only about nine percent of servers run virtualisation software, together with a miniscule number of desktops.
Yet there's not a great deal of animus against VMware - apart from its competitors that is, who all complain bitterly about it. But then, they would, wouldn't they?
Rather, enterprises have bought into VMware's technology pretty much by default. Of course, it doesn't have the market to itself but no competitor has looked close to disturbing the Palo Alto-based company's smooth cruise to IPO and beyond.
Its US$28 billion market cap looks excessive for a company whose quarterly revenues were US358 million for the last reported quarter in October; there's a fourth-quarter report due on 28 January. But the background is that the figure was a chunky 90 percent up on the year before.
Post-IPO, VMware has used its cash to cement its market position with new products. As well as the hypervisor itself in the shape of ESX Server - which analysts generally agree will become free in the near- to medium-term future - VMware has a fairly strong, enterprise-focused product set surrounding its virtualisation technology.
The focus now is on datacentre management via VMotion and associated technologies, as well as test and development tools such as Workstation, Lab Manager and Stage Manager. And there's a range of third parties - VMware loves the ecosystem metaphor - such as Platespin, bringing applications such as high availability and disaster recovery.
The rumble of Redmond tanks
But Microsoft is coming. And it showed its intent in no uncertain terms with the recent acquisition of Calista Technologies, announced this week.
Calista's technology is all about the desktop, an area on which Microsoft, for all its enterprise protestations, is still firmly rooted. One Microsoft insider reckoned that the ethos at Redmond is still tilted towards ensuring the "best end user experience." In other words, that's what back-end systems are for; he wasn't the first to say that and I doubt if he'll be the last.
Microsoft bought Calista for its virtual GPU technology, called Calista Virtual Desktop (CVD). It's claimed to boost the performance of graphics-intensive applications, whose high-bandwidth demands have long been the constraining factor for those wishing to deploy remote desktop systems. It takes a lot of bandwidth and low latency to push seamless graphics, such as the Vista UI, over a network.
It's an article of faith within Fortress Redmond that the desktop is Microsoft territory, and Calista is a key element in furthering that strategy.
But Microsoft is pulling together the pieces it needs to present a full portfolio to the market. It has announced new, more liberal licensing terms for Vista when run in a VM. This is true for home buyers of Vista Home and for the enterprise, where VECD - Vista Enterprise Centralised Desktop - allows you to run Vista in a VM for only $23 per desktop under its Software Assurance package. According to Microsoft, it's essentially a licensing enabler which saves you buying the full packaged product.
Microsoft also has Hyper-V which, if Microsoft gets it right, will at some future point be a true competitor for VMware's mature hypervisor technology.
More importantly, since the hypervisor is tending towards freebie status, it has all the management pieces too. With System Center as the über-manager, its Virtual Machine Manager product offers control over VM provisioning and general management, using SQL Server for its database, while the clustering services embedded in Enterprise Edition of Windows Server 2008 provide availability and backup. According to Microsoft, under Windows Server 2008's VM management, migrating a VM in the case of a planned outage takes between five and 60 seconds, depending on the infrastructure and application.
The Citrix factor
Some argue that Microsoft's offerings are not as mature or as comprehensive as VMware's. However, Microsoft is also drawing on Citrix's expertise as a result of the controversial technology-sharing agreement that Microsoft and XenSource drew up before the latter's acquisition.
Microsoft and Citrix share a common VM file format in VHD and have just announced a new, more comprehensive partnership. This means you'll be able to manage Microsoft VMs using Citrix/XenSource's tools - and XenSource has been working on this technology for some eight years so its technology is more mature than Microsoft's.
Microsoft says the new Citrix deal means it can add interoperability -- the company called it "acting responsibly". Incidentally, it's not quite as clear exactly what positives Citrix pulls out of this deal, since it looks increasingly like it's being boxed in by Microsoft and VMware. But the Citrix guys seem happy about it, so hey....
Microsoft is also ramping up the noise levels on its application virtualisation product. Dubbed -- in Microsoft's typically anodyne fashion -- Application Virtualisation, it is nakedly near-identical to the SoftGrid product Microsoft inherited from its Softricity acquisition back in 2006; Microsoft has been working on this strategy for a long time. It sees application virtualisation as a way out of DLL and registry hell, and as a way of speeding up older installations of Windows, which get bogged down with vast numbers of registry entries -- many of them defunct. It also gives IT staff more granular control over which applications and which application feature users can run.
As you might expect, Microsoft is pretty bullish about its approach to the virtualisation space, its technology, and its ability to pull together the pieces needed for a concerted attack on the established market leader. "The technology needs to sit within a platform and ecosystem of partners and skills that understand it," said a spokesman. "It's the ability to exploit the whole platform that's significant so that you have multiple options to deliver a desktop to the user within our ecosystem."
Microsoft is after all executing the strategy it outlined in May 2006. In particular it sees its desktop strategy as the Trojan horse, a way of getting a foot in the door. It supports remote Windows and Linux desktops via RDP, the Citrix-developed, remote desktop protocol that's built into Windows mainly for the purpose of providing remote technical support. This "directly addresses the two major barriers to widespread virtual desktop adoption: the quality of the user experience and the cost per desktop," said a spokesman.
Microsoft is not alone in its rosy view. IBM's VP of worldwide server services agrees that the companies have a good chance of rolling off VMware's lawn and into its HQ. Ready should know - he heads up the chunk of Big Blue that's responsible for selling server consolidation services - read virtualisation - into global corporations.
But for Ready the issue is whether Microsoft can execute - whether it will follow its usual course and get it right after several attempts. "Look at the browser market," he said. His recipe for VMware is either acquire or develop more management tools.
So VMware's prospects appear to be good in the short term, and probably into the medium term too. After all, with so small a number of servers virtualised, there's plenty of room for expansion. But as long as VMware remains a significant player, there'll be no legislative bodies stepping in to accuse Microsoft of monopolistic practices. And if Calista's technology works as claimed, then it suggests that VMware, which has always focused first on the back end, needs to sharpen its act with respect to the corporate desktop.
Despite that, Microsoft does have several advantages up its sleeve, not the least of which are its determination and tenacity, its stranglehold on the OS, its experience in winning from a position of zero market share - and its deep, deep pockets.
You'd have be brave to bet on Microsoft losing this fight in the long term - but, taking a very long shot, if IBM were to buy VMware, maybe the outcome will be different.