Does *free* virtualization = certain chaos?
Round-up of discussion from week two
Reader workshop This week we broached the subject of 'stuff that can go wrong with virtualization'. One of the comments we received on last week’s round-up article serves as a good lead in. The Reg reader was asking for:
'free, good virtualization software [to] come with server OS'es as standard...once virtualization software costs drop to "zero" and you don't have to involve a third party to virtualize, you'll see lot more companies consider it.'
Well, the good news is that things are certainly moving in this direction. Virtualization capabilities are provided as standard by Red Hat and other Linux vendors for example. Meanwhile, the token fee Microsoft initially added to Windows Server 2008 for the hypervisor has disappeared.
Does a mainstream vendor selling 'server + virtualization' as a single package mean we're approaching the point of 'zero cost' virtualization? Building on this theme, the notion of *everyone* doing virtualization does not appear completely alien. But as we reported, skills sets are not always keeping up. Does this mean a deluge of problems caused by virtual-newbies, or are there enough IT pros to go round who have already cut their teeth?
The availability of skills was one of the concerns raised this week:
Some [small shops] fall by having only one guy trained in the underlying technologies… In a physical world, you expect a server support tech to understand the hardware and OS an app resides on. In virtual environments, server support techs looking after VMs should also have at least a basic knowledge of the hypervisor layer. Equally, I've seen large shops send a few people on a training course and expect them to design, implement and migrate onto a virtualization platform without any prior experience.
That doesn't sound like a recipe for getting to grips with virtualization, or indeed any IT related initiative. But does it really matter, or is the very point of getting to grips with virtualization the ability to create VMs, make mistakes, start over and so forth? We would suggest that the answer is a voluble ‘no’ – but one of the downsides of virtualization is (as one canny reader pointed) the risks of putting one’s eggs into a single basket. In other words, the very fact that multiple systems are being hosted on a single box, means that operators need to be even more highly skilled than before:
The main danger posed [...] is not understanding the risks. Running 10 VMs on one host means that if (when?) that single piece of hardware blows a fuse then you haven't just lost 1 service, you've lost a whole bunch - so the hardware reliability drops by a factor of the number of VMs each box is hosting.
The easy answer here, of course, is that anyone worth their salt wouldn’t consider virtualization without appropriate disaster recovery and back up in place. However, anecdotal feedback from previous Reg studies in that particular domain suggested that many IT shops - often through no fault of their own – have had to endure major failures before the DR side of the IT equation is taken seriously at the business level.
The notion of hidden risks was a clear winner in this week's comments. What's interesting perhaps is how these are not unknown issues. However, they are issues that are rendered more complex when virtualization is added to the mix:
Firstly there's config management, especially insofar as it affects software licensing, management, performance and capacity planning. If you are going to move your apps all over an ESX farm you had better have a way of dealing with all those issues... Then there is the support problem - I've lost count of the number of suppliers that don't support virtualized environments.
In some cases, virtualization may indeed make things harder rather than simpler. Indeed, a newly initiated IT shop could be forgiven for thinking they had in fact swapped a lot of traditional problems for a whole bunch of new ones, the implications of which are yet to be fully understood. We can extend these thoughts into one specific area of risk, i.e. IT security. In this, as with other areas, the stock answer would appear to be due diligence.
Perhaps the only guidance that can really be given at this stage is around due diligence – at the heart of security best practice is the eyes-wide-open mindset, in which risks are clearly understood and appropriately dealt with.
This does however lead us back to the question about skills. Perhaps the biggest risk of all involves the potential performance and security problems caused by the people not knowing what they didn’t know. There are no hard and fast answers yet, but needless to say, we’re keen to hear your side of the story on these topics, and indeed any other topics you’d like to bring up, to keep the conversation moving and to set the scene for next week.
I'll leave you with my favourite excerpt from one of this week's comments:
'...Proper planning, management and implementation negate the vast majority of the problems raised’. We wager most of the IT pros out there will heartily 'Amen' that sentiment and do their utmost to apply it in practice. But we’re not yet at the stage where we know what a proper virtualization plan looks like, never mind implement it.