|
 |
|
General Information
|
Add To My Personal Library |
February 12, 2010
• Vol.32 Issue 4
Page(s) 20 in print issue
|
Watch Out For Virtual Sprawl
Take Steps To Keep Your Virtual Servers In Check
|
Over the last few years, virtual environments have risen in popularity at an unprecedented rate. VMs (virtual machines) are being used for everything from hardware consolidation to energy efficiency. Problems arise, however, when virtual machines are improperly implemented or managed.
“Virtual sprawl is the unmanaged proliferation of virtual machines caused by users cloning virtual machines,” says Ido Sarig, chief evangelist and vice president of marketing for BDNA (www.bdna.com). “Virtual sprawl can have serious implications—from management headaches to litigation risk, as virtualization spawns unknown and unaccounted for copies of licensed software.”
So what does it take to avoid virtual server sprawl—or to rein it in once it has taken root?
 Inventory Everything
The first step in resolving virtual sprawl is to inventory everything. Manually trying to account for all assets in a virtual world is all but impossible. Therefore, it is best to use an IT asset discovery tool to identify all hardware and software in existence, whether in a virtual or physical state.
“IT organizations need to implement automated solutions—and ensure that the asset inventory tools they select are capable of discovering what software is installed on those VMs, and whether or not any of it is used,” Sarig says. “It isn’t enough to determine that a particular physical server is running several VMs.”
Don’t Solve One Problem By Creating Another
SMEs can often avoid virtual sprawl by paying attention to what’s going on with the VMs. A common issue relates to improperly fixing one application, which then leads to another app becoming inoperative. This can come about simply by changing the settings. In the quest to consolidate and virtualize everything, make sure new problems are minimized.
Greg Schulz, an analyst with StorageIO Group (www.storageio.com), explains it in the context of a redesign of backup and data protection applications.
“The way to avoid something breaking is to go in and look at what VMs along with guest operating systems and associated applications can be consolidated,” Schulz says. “Pay attention to what other applications will be affected, including backup and DR. Think of it as the first step or phase to consolidate physical machines (PMs) into VMs on fewer PMs.”
Create A VM Checklist
The great thing about virtualization is how easy it is to create virtual servers. That speed, however, is one of the big contributors to sprawl. People just don’t pay enough attention to what is going where. Paul Schaapman, a solutions architect at CDW (www.cdw.com), says the remedy is to create a checklist designed to ensure that new servers are only created based on real need.
Such a checklist should include details of the business requester; a description of the business need; the application to be deployed; the resource owner; the business unit it falls under; application support teams that might be impacted; the targeted end date for use of the server; and the memory, CPU, and space requirements, including the backup and storage requirements. This will help ensure that new VMs are only created when actually needed.
“Once this is completed, IT can assign a cost to the new virtual server,” Schaapman says. “These costs would include operation system licenses, client access licensing, as well as space, backup, and DR costs. The total cost picture should be approved by the appropriate parties.”
Marry Virtual Servers With Virtual Storage
When planned efficiently and managed well, virtualization can increase the utilization of servers from a rate of 15% to as high as 85% while at the same time enabling high availability and disaster recovery for applications. On the downside, though, Chris McCall, product manager for HP’s Unified Storage Division (www.hp.com), points out that virtualization demands more from the network and storage infrastructure, which can lead to performance bottlenecks and make it harder to recover from a site failure or disaster.
“Because virtualization depends on shared storage to implement disaster recovery and high availability, customers need to leverage flexible, feature-rich, virtualized storage solutions that can quickly adapt to their business needs,” McCall says.
Such solutions leverage purpose-built systems to create a pool of virtualized, shared resources that can significantly improve utilization of hardware, increase scalability, and drive flexibility across the infrastructure. This extends virtualization capabilities across the entire hardware infrastructure via pools of available capacity, bandwidth, and CPU that allow capacity to be quickly assigned or repurposed on the fly to support various critical business applications. As a result, the SME can respond more rapidly and make appropriate changes. The virtualization of storage can also lead to more efficient data provisioning, data reduction, and data placement. With that in place, VMs are better supported and management can be simplified, as performance and availability roadblocks crop up far less often. 
by Drew Robb
BONUS TIPS
Use agentless asset discovery tools. Agent-based solutions require an installation of agents on physical servers, which might not play well in a virtual environment and can make the agents become a part of the problem. After all, they need to be maintained, upgraded, and managed. Ido Sarig, chief evangelist and vice president of marketing for BDNA (www.bdna.com), suggests an agentless solution to asset discovery as the best approach. Otherwise, you may end up with rogue physical servers sitting around the data centers running VMs you might never find out about, as no one knew to install an agent on those boxes.
Manage all facets of the virtual environment. There are many facets to effectively managing a virtual environment. According to Paul Schaapman, a solutions architect at CDW (www.cdw.com), these include using a hypervisor patching process to keep the different operating systems up-to-date as well as a server patching process to keep servers up-to-date. Backup, power management, and virus scanning are just a few examples of the various areas that have to be given attention. |
Cautionary Tip: Beware Of Licensing Problems
Licensing hassles, those banes of the IT manager’s existence, become even more of a pain when you start dealing with virtual machines. Hosting applications on a VM can open up a Pandora’s box of licensing evils. In many cases, every VM on the physical server needs its own license to avoid license noncompliance. Yet some go on copying software onto new VMs, blissfully unaware of the licensing consequences. Watch out, therefore, for such practices as having a “master” VM containing licensed software, which is then copied onto every new server.
“The deployment of software in virtual environments adds several layers of increased complexity to an already difficult task of managing license compliance,” says Ido Sarig, chief evangelist and vice president of marketing for BDNA (www.bdna.com). “In many cases, companies don’t have policies in place to control and monitor the deployment of software onto VMs.” |
Smart Tip: Consolidate VMs
Just as you can have too many servers, similarly you can have too many VMs running, which can lead to poor levels of utilization. Underutilized VMs are harder to manage and easier to lose track of, plus they often consume a lot more resources than necessary.
“Take a close look to see what applications or systems can be further consolidated,” says Greg Schulz, an analyst with StorageIO Group (www.storageio.com). “For example, turn 10 poorly utilized VMs into four that are properly utilized.” |
|
|
