What we call today server virtualization (hardware virtualization sometimes) is a compelling technology, quickly pervading worldwide datacenters with the promise of money saving and more efficient management.
While this is true, there are a lot of factors which concurrently reduce benefits of server virtualization: high starting costs for hardware, complexity and discrepancies in software licensing, lack of support, lack of experienced professionals and training, lack of tools able to address long-term issues.
Server virtualization is a revolutionary technology for sure, but it’s still very immature in several aspects: some of the above shortcomings will drive future market trends.
First of all SANs will become a mandatory companion for virtualization, obliging even smaller companies to purchase network storage equipment, falling back to cheaper NAS only for very small projects.
To reduce costs and be ready to scale up, cheaper iSCSI models with modular architectures will be the preferred choice.
At the same time virtualization will drive sales of high-density multi-core CPUs, which leads to higher consolidation ratios.
A single eight-core host will easily accommodate 32 virtual machines on average, which is more than enough to build a complete datacenter for many SMBs. And if this is not enough Intel is already working on an 80-cores prototype, available in production within five years.
In such scenario hardware will have to be more redundant and reliable than ever, obliging customers to buy more expensive hosts, with each physical component doubled, including motherboard and CPU.
Concurrently vendors will start to satisfy a moderate demand for enhanced disaster recovery capabilities in virtualization platforms from enterprise customers having to address severe high availability requirements.
Above the hardware, companies will start preferring solutions which are already supported in virtualization scenarios, urging software houses to reconsider their support and licensing models.
As soon as application virtualization will become more popular the already started evolution of licenses will have to change its path again, probably finding that usage tracking of physical resources is the best way to go.
Anyway the process is expected to take a lot of time, with almost no chances to successfully complete within this year.
Going further on software side, the most impellent issue customers will have to address will be management of hundreds, even thousands of virtual machines, with or without licensing revolution.
Here vendors offering datacenter automation tools have the chance to collect a notable success among enterprises starting from end of 2007.
But server virtualization is not the only technology doomed to success.
On the client side I think application virtualization market will start growing in the second half of the year, after Microsoft, Citrix and Symantec will revolution application distribution with products acquired during 2006.
A last note: the battle for market dominance, actually seeing VMware, Microsoft and XenSource/Virtual Iron as main contenders, will start after launch of new Windows hypervisor, codename Viridian, planned somewhere between end of 2007 and beginning of 2008, and will play entirely on the SMB market, where initial startup price is the biggest obstacle in virtualization adoption.
So expect a remarkable attention to smaller companies in the coming months.
This article originally appeared on SearchServerVirtualization.
