Can virtualisation reduce the management overhead?
Yes and no. Maybe
Lab Server virtualisation is quickly becoming a mainstream route by which IT services are delivered. Already much of the “low hanging fruit” has been identified, with server consolidation projects providing an entry point for virtualisation. But after such implementation projects are completed the challenge will become one of managing virtualised systems in daily operations. Does virtualisation make living with IT systems simpler or more difficult?
We know from research with Register readers that virtualisation, especially of server systems, is more widely deployed than in the past. However it is becoming clear that whilst some of the obvious benefits expected from such systems are being enjoyed, challenges are also becoming apparent.
The research shows that the majority of IT professionals are positive about what virtualization can deliver in terms of rationalizing x86 Windows and Linux server estates. However, the data also points out that there are issues to be addressed in the design and configuration of virtualised environments. Furthermore, and much more importantly, it is clear that identifying operational best practice and using such in everyday operations are matters with room for improvement.
The ease of creating new virtual machines is resulting in an ever larger number of virtual machines being run. The ease of starting new virtual machines can be beneficial in many scenarios, notably by simplifying the often complex management workload associated with providing rapid disaster recovery or service resumption following ‘faliure’. Clearly this is a very good example of management operations being enhanced with virtualisation.
On the flip side though, experience shows that this virtual machine sprawl can also have negative effects on previously routine management activities. Amongst the areas identified as posing problems in virtualised environments is the modern take on patch management, where organisations are still seeking effective tools and procedures to patch virtual servers, especially when they are dormant.
Other similar issues are creating operational challenges. Vendors shout about how quickly new virtual systems can be spun up if service response drops or workloads increase, but this raises the question of how should virtual machines be monitored for health and service performance? With the creation of virtual machines becoming a simple task - or even subject to automation engines - do you have processes in place to ensure that application and other software licenses are available to ensure that systems are operated in accordance with established license terms and conditions?
Clearly keeping virtual server proliferation under control is a key element in managing virtual systems. As well as demanding the use of good tools and establishing sound operational working practices, this will also mean that some internal political and, perhaps, budgetary processes need to be updated to reflect the new IT infrastructure. We know that many budget holders still want “their physical systems” to host only their virtual applications rather than sharing the physical resources amongst multiple budget holders.
So, the key question becomes – does virtualisation make management easier or harder? We suspect neither, for the simple reason that any extra management capacity made available by consolidation will be more than taken up by subsequent proliferation of virtual machines. This may be an anathema to the virtualisation evangelists, but what is clear is that management tools and (more importantly) processes to deal with virtualised environments are thus far lagging behind those available in the physical world.
As things move forward, based on reader feedback among others, it would appear that the operational administration of virtualised systems will need new management processes to be identified and refined – and these will need to be supplemented by new management tools. How are things evolving in your organisation? If you have any feedback, please drop us a line in the comments box below. ®
Sponsored: Hyper-scale data management