A recent study by Cisco Systems, NetApp, and VMware surveyed IT managers, directors, and CIOs in the small to medium-sized business market about virtualization adoption. (Small businesses are those with 50 to 100 employees, and medium-sized businesses are those with 100 to 500 employees.) The study found that 65% of small businesses and 79% of medium-sized businesses had adopted some form of virtualization.
For those considering virtualization, the greatest hindrance is cost. Interestingly, it's middle managers who are most skeptical about costs, with 61% expressing concerns. In contrast, just 33% of upper management listed cost as the main hindrance to adopting virtualization.
Also, 91% of survey respondents who implemented some form of virtualization believed their companies had competitive advantages over those without virtualization. Even 71% of respondents from companies without any virtualization in place thought it could give them an edge over the competition.
Dealing with funding
Mark Oliver_Head Shot_GroupOliver_111913.jpg
"Because of that, some larger vendors such as Microsoft and Cisco now offer competitive financing plans and will finance the entire package including labor and software," he said. "Currently, Cisco has three-year 3% and a 90-day delay payment programs."
Brock Jamison, vice president of sales at Orion Networks, suggests that companies consider their financing options well before they start down the virtualization road.
"Right now, small businesses are in luck as rates for financing technology upgrades are at historic lows," Jamison said. "However, this isn't going to last forever. As most businesses will eventually adopt some form of virtualization in the near future, it’s smart for businesses to explore their financing options sooner rather than later."
Using policies to address server sprawl
Virtualization has its own list of potential problems and chief among those most often cited is server sprawl. However, Oliver said that issue can be addressed with policies.
"Of course there is software that can help, but server sprawl is a symptom of more fundamental problems in a virtualized environment," Oliver said. "Core issues are handled by management policies. Policies or procedures need to include who can deploy servers, how they are maintained, who maintains them, and who decommissions servers. Additionally, policies for naming servers, where they reside, how resources are allocated and the like, also need to be addressed. Without some basic policies, things can get out of control and create similar management issues that were part of the reason why virtualization was adopted."
Jamison, too, said planning was key to keeping sprawl at bay.
"While virtualization offers small businesses loads of benefits, it can also be difficult to manage," Jamison said. "This coupled with the relative ease in deployment and the tendency to ignore and mismanage a server once it's deployed makes server sprawl a common problem. However, by following these key steps, ones that are often overlooked during the planning stage, businesses can further prevent becoming a victim of server sprawl."
Those steps include succinctly assessing organizational needs, determining the length the solution is required, and properly documenting the deployed systems to enable proper future management.
Keeping the NICs in order
Another challenge that has been associated with virtualization is incidences of network interface cards (NICs) getting saturated and then reporting network errors.
"There are a few things to keep in mind, or ways to help," Oliver advised. "Initially, start by making sure to get enough physical connections on the host machine. When configuring, give the high volume activities a dedicated physical connection through the switch infrastructure. When things are operational, use bandwidth monitoring to keep track or tabs on traffic. The long and short of monitoring is your two main options are SNMP and Promiscuous monitoring. One can use a bandwidth shaper like NetLimiter as well."
It's still about the hardware
Many organizations look to virtualization to solve network downtime, but Oliver said there still needs to be solid hardware in the system.
"Disruptions can be alleviated or minimized through the use of redundant fans, power supplies, and the like," he said. "Additionally, drive configuration that provides for backup or redundant physical drives such as RAID 1, 5, or 10, depending on the system demand. Having a good support contract typically from the manufacturer also helps. Following the solid hardware guideline, having a good backup and restore solution that allows for fast replication or recovery is needed. Also, establish a host server and SAN hardware consolidation. Then, as storage space allows, use the host server storage instead of full SAN, which can expedite the speed of recovery."
Doing advanced testing of existing applications is a key aspect of adopting virtualization, and Oliver offered advice based upon the type of applications. When doing a typical virtualization by migrating existing services or applications from an existing physical server to a virtual instance, he said the level or degree of testing is largely dependent on how critical the service is to the organization. For less critical applications, it is typically done during off-hours, and testing in production with the knowledge that one can go back to the original server if things are not working as expected. For more complicated or critical services, typically the existing system remains online, while offline the virtual machine is established with the service, and connected to a test database as required and then tested for both functionality and performance. Functionality and performance testing can be done manually or with tools if available.
Expert help for licensing
Software licensing issues can also be problematic in virtualized environments, and Oliver stresses the need for expert help.
"Software licensing rules are a challenge under the best circumstances," he said. "Virtualizing licensing is difficult. It is best to talk to a licensing specialist that is associated with your software vendor, as well as have a solid understanding of your infrastructure and how users and devices are communicating with the applications or services. For example, with Microsoft Virtual Desktops, it makes a difference if the user or device connecting to the virtual desktop is a thin client or desktop, and if the desktop has an operating system covered under a Software Assurance. Conversely, some databases can be, or are licensed by the type of processor it uses. Further complicating things, there are some things you can't do with OEM or retail licensing, but one can get it covered with additional service."
The other instance that can make things a challenge, he said, is how the remote, backup, or disaster recovery software is licensed. Some manufactures consider a remote or disaster recovery instance to be "warm" and running; therefore, it requires a full license. Other manufactures allow for this with only the production copy, and yet others have a special and typically discounted version for the warm copy. The short story is that licensing is complicated, and frequently a specialist is needed to help.
Duane Craig reports and writes on technology, construction, finance, food, and agriculture. He's been published in trade print magazines, the Washington Post, and widely on the web.