The emergence of powerful virtual machine software for industry standard (X86) systems has changed how many organizations look at deploying server-based applications and services. In the past, many organizations would deploy a single workload on a server. While this approach might have been easy to understand, it was wasteful of system resources.
Virtual machine software, or a hypervisor, from suppliers such as VMware’s ESX, Citrix’s XenServer, Microsoft’s Hyper-V and both KVM and Xen from the open source community has caused IT architects to think differently about deploying workloads and services. It is now common to encapsulate a workload or a service into a virtual machine and execute multiple workloads or services on a single X86 server. In essence, IT has begun to think of system resources the way mainframe developers have for years.
Sever virtualization, while helpful for workload consolidation and system optimization, brings along with it a number of issues as well:
Ease of creation — the creation of new servers is now extremely easy and can be accomplished in minutes or hours rather than days or weeks as in the past. It is now so easy that IT administrators may not have any idea that a business unit or department has created a new virtual server.
Since many organizations don’t have the proper tools to see what is happening in the data center from moment to moment, IT administrators are hard pressed to assert guidelines and make sure that they are followed. This can lead to organizations running unlicensed software. It also leads to virtual machine sprawl.
Monitoring Must Be Real-Time
The tools simply must present the data in an easy-to-understand fashion that shows how all systems are running. They must support all of the commonly used operating systems on these physical or virtual industry standard systems.
One of the more challenging issues IT faces in today’s complex physical/virtual/cloud environment is knowing what is happening, where it is happening, what resources are being consumed and who is responsible for consuming those resources. When a performance problem appears, it is often an unwelcome, difficult to diagnose surprise.
Determining what is responsible for a workload slow down is very challenging because the workload may actually be spread over multiple tiers of servers (physical, virtual or somewhere in the cloud). When a slow down occurs, IT may not be able to determine simple facts such as the following in time to head off an outage.
- Is this a network outage?
- Is the database performing badly? Is it doing so because of a database error or a storage system error?
- Where is each tier running? Is it a physical machine, a virtual machine or a virtual machine running out in the cloud somewhere?
It is very important for IT to be able to find and resolve the root cause of the problem rather than simply chasing after symptoms created by the problem.
Analytics and Optimization
To be successful, today’s IT organization must use tools that find resources without help, collect run-time data without having a negative impact on workload performance and provide a quick analysis of what is happening. The best tools would go on from there to suggest ways to optimize the environment and prevent problems in the future.
Regardless of the time of day, IT must know what's running, where it is running, and what issues are emerging long before they become problems. IT monitoring tools, such as those offered by Zenoss, that offer the capability to gather data, analyze that data and present a succinct and yet comprehensive view of what is happening are a requirement to manage today’s complex data center.