Time to Get Real with Virtualization Management

Arthur Cole

The steady drumbeat of opinion regarding the need for virtualization management has not had a dramatic effect on data center managers so far, judging by the latest survey results. But as the number of virtual machines continues to grow exponentially, it probably won't be long before reality demonstrates what informed opinion could not.


That is, enterprises across the spectrum are badly in need not just of virtual management systems, but unified, hybrid solutions that span resources and applications across physical, virtual and cloud platforms.


To be fair, it's not like the top virtual platform vendors have been leading the charge as far as management goes. VMware, Microsoft and Citrix have only recently stepped up to the plate with what could be described as effective management platforms. But it seems even they are having trouble keeping up with the rapid pace of virtual/cloud deployments out in the field.


That's why many enterprises are turning to outside consultants to find a way out of the virtual messes they've created, according to Yeoman Technology Group's Michael Healey. When there are only a few hundred VMs humming along, the problem of management seems insignificant. But before you know it, there are thousands, with little or no record of who created them, whether they are still being used, and whether they can be decommissioned without disrupting critical workflows.


Even when they do get serious about virtual management, most organizations tend to focus exclusively on virtual sprawl, according to Embotics' David M. Lynch. But it's important to remember that sprawl is a symptom of poor management, not the cause. Only by going after root problems, like the lack of system insight and reporting, administrative lapses and the absence of an effective automation regime can there be any hope of getting a handle on an out-of-control virtual environment.


That's part of the reason why visibility into the virtual world will be a crucial area of development in the coming year, says Stratus CIO Joe Graves. This need will be complicated by the arrival of multi-vendor virtual environments and the fact that they will be increasingly tasked with mission-critical applications. Establishing a single management infrastructure that can handle these complex environments, preferably without having to tear out existing management systems in the process, will be the single-most significant challenge if enterprises hope to extend their current virtualization practices.


It would be easy to say that effective virtual management is just a matter of plugging in the right tools and letting the system take care of itself. But adding another layer of abstraction onto the enterprise infrastructure means that an entirely new approach to management is needed -- one that encompasses hardware, software, systems, and the policies and procedures governing their interaction.


And as one would expect, it starts by knowing what you have on the ground already, and what you hope to achieve through virtualization.



Add Comment      Leave a comment on this blog post
Feb 17, 2010 2:38 AM Tyler Lane Tyler Lane  says:

This is such an important point for all of IT and data center managers as well. There are several directions for virtual sprawl to go, with the easiest likely to become the most common: ignore/restore. In some ways the benefit to virtual machines allows for multi-layer redundancy (i.e. cloud), and therefore one could argue that the physical location of a virtual server should not matter. In this context, we might have virtual machines that simply spawn another instance to support higher throughput and, when a server, be it physical or virtual fails, the application running on it could be picked up by a backup - without service disruption. If downtime is not a factor then IT might be happy, in the short term.

However, if one were to focus on the data center, the eventuality is chaos; when to replace hardware when you do not know what is running on that hardware. Data Center Service Management (DCSM) is one potential bridge over that gap. With DCSM tools and practices in place the data center will be better prepared to bend in the direction of the week, month or year.  As Joe Graves points out "visibility into the virtual world will be a crucial area", and for the data center it may be an opportunity to re-tool and become a bigger, more integral part of enterprise architecture.

In some ways the impending issues - real, not virtual - are similar to issues with an Enterprise Service Bus (ESB), where many services are configured for maximum accessibility and throughput. The virtual server is essentially equivalent to "a service", or one might call it a "meta-service" as it may provide access to multiple services, But when SOA was introduced, into or parallel to the ESB the result was often more confusion, with less security, rather than greater access with higher uptime.

Virtual servers are here now and many more are coming. Plan your method of attack and coverage today!

Reply

Post a comment

 

 

 

 


(Maximum characters: 1200). You have 1200 characters left.

 

null
null

 

Subscribe to our Newsletters

Sign up now and get the best business technology insights direct to your inbox.