Why server virtualisation can store up trouble

The cost savings and performance offered by server virtualisation can raise unforeseen issues for storage and the wider infrastructure, says research analyst Martin Atherton.

Server virtualisation technology is taking hold in mainstream IT environments and is on its way to becoming a core technology. But the views I hear time and again from IT professionals are as much about virtualisation’s impact on the wider infrastructure as about their experiences of the technology itself.

One of the side effects of virtualisation is its impact on the storage infrastructure. To understand this issue, we need a grasp of how IT environments have traditionally evolved. Generally, each time a new application was deployed, a new set of servers and storage devices came with it, creating a so-called silo.

This approach causes some problems: it requires physical space for a start, and managing multiple, disparate hardware and software platforms has proved onerous for many organisations. Nonetheless, it has one major benefit in that each application’s storage requirements are handled more or less individually.

The paths between application processing and storage, and the bandwidth required, can be calculated in a relatively straightforward manner for individual applications. While bottlenecks have always existed, the nature of silos makes them relatively easy to spot.

But there is an additional downside. Because storage systems have been fully specified upfront to cope with expected loads, much spare capacity exists today that has never been used but which is unavailable for anything else.

Enter server virtualisation, which offers a way to consolidate multiple workloads onto a single physical server. The benefits are generally well appreciated and in principle can also result in more efficient use of storage by breaking the silos that were bought to serve the needs of particular applications.

However, the downside of server virtualisation is that it doesn’t pay sufficient attention to the physical environment beyond the server. It’s great in principle to shift multiple workloads onto a single server. But rather than having individual routes to the physical storage layer, information flows now take place using the same physical interface.

From talking to early adopters of server virtualisation, we know this approach can easily cause a bottleneck if the workloads are data-intensive or require access to concurrent physical resources. Where silos make bottlenecks easier to spot, virtualisation does not.

Numerous IT managers describe their difficulties understanding what physical resources a virtual machine is actually using. Management tools – or at least those used by many IT departments – are not sophisticated enough to provide such a view, and many IT staff lack experience of the new environments.

However, we should also remember it is still relatively early days for virtualisation in the distributed systems environment. IT management vendors are working with virtualisation vendors to ensure the physical and virtual worlds are both taken into account.

Meanwhile, although the traditional storage infrastructure of NAS, SAN and the like evolved without considering virtualisation, a number of vendors are starting to incorporate recognition of server virtualisation in their products.

In fact, the principles of virtualisation are also highly applicable to the storage realm and happen to have been around for a long time. Indeed, storage virtualisation makes the overall pool of storage easier to manage and enables the allocation of storage as it is needed through so-called thin provisioning, reciprocating the operational flexibility of running a virtualised server environment.

Storage virtualisation also offers a number of other benefits such as reduced power consumption and makes supporting disaster recovery easier, as data can be replicated and moved within the overall storage pool with minimum disruption to users and applications.

Ultimately, one of the key things we’ve learned from early adopters is that it doesn’t make sense to think about server virtualisation and storage virtualisation in isolation from one another. Looking at virtualisation across both areas should improve performance and efficiency, and cut costs.

Storage is often an area that hasn’t been revisited for some time in many organisations, so it makes sense to study how the latest capabilities can support server virtualisation, before operational bottlenecks threaten to spoil the fun.

Content Contributors: Martin Atherton