The fractal data center

In a recent newsletter we introduced the concept of Application Delivery 2.0, a major component of which is virtualization. This is the third in a series of newsletters that discusses how we think that virtualization will dramatically shape 2010 and beyond. This newsletter will focus on the management challenges associated with the ongoing virtualization of data center servers.

The life and death of a virtual machine

The last two newsletters talked about the movement to implement virtualized data center servers. One way to think about the current generation of virtualized data centers, and the related management challenges, draws on the concept of a fractal. A fractal is a geometric object that is similar to itself on all scales. If you take a high level view of a fractal and observe its shape and if you then zoom in on the object, what you see will look similar or exactly like the original shape. This property is often referred to as self-similarity.

The relevance of fractals is that the traditional data center is comprised of myriad physical devices including servers, LAN switches and firewalls. The virtualized data centers that most IT organizations are in the process of implementing are still comprised of physical servers, LAN switches and firewalls. In addition, these data centers also house virtualized servers which themselves are comprised of myriad virtualized devices including virtual machines (VM), a virtual LAN switch and in many cases virtual firewalls. Hence, if you take a broad overview of the data center you see certain key pieces of functionality. If you were to then zoom inside of a virtualized data center server you would see most of that same functionality – hence, a fractal data center.

In a fractal data center, IT organizations still need to perform all of the traditional management functions on the physical devices in the data center as they always have. In addition, they now need to perform those same functions on each VM. For example, just as IT organizations need the ability to discover physical servers, they now need the ability to discover VMs.

However, performing a traditional management function such as troubleshooting in a virtualized environment can be extremely difficult. For example, assume that an IT organization is running an n-Tier application in a virtualized environment and that the Web server, application server and database servers are running on VMs that are housed in the same physical server. Further assume that the users of this application complain of sporadic poor performance.

Unfortunately, we are still in an environment in which the assumption is typically made that the WAN is the cause of any degraded application performance. Once the network organization has proven its innocence, which could take days, it is likely to take additional days for the IT organization to discover that the reason that the application performs poorly is that the application server is experiencing sporadic performance problems. In order to make this determination, the IT organization needs detailed information on each of the three VMs and the communications between them. Part of the challenge associated with this scenario is that in most cases once an IT organization has virtualized its servers it loses insight into the inter-VM traffic that occurs within a physical server.

However, just determining that it is the application server that is causing the application to perform badly is not enough. The IT organization also needs to understand why the application server is experiencing sporadic performance problems. The answer to that question might be that other VMs on the same physical server as the application server are sporadically consuming resources needed by the application server.

Solving this type of problem will require a level of management data that most IT organizations do not currently collect. Yet not being able to solve this problem at all, or not in a reasonable timeframe says that IT organizations are taking a major step backwards relative to being able to ensure acceptable application performance.

In the next newsletter we will discuss desktop virtualization, a topic that we talk more about in a recent report. In the mean time, we would like to hear from you. What forms of virtualization have you implemented and what new challenges did that create for you?

Learn more about this topic

Half of new servers are virtualized, survey finds

The life and death of a virtual machine

Virtualization: Hot technology for 2009

Join the Network World communities on Facebook and LinkedIn to comment on topics that are top of mind.

Copyright © 2010 IDG Communications, Inc.