Challenges in Virtualization

By Sue Poremba

Sue Poremba is a freelance writer focusing primarily on security and technology issues and occasionally blogs for Rackspace Hosting.

Virtualization has been a boon to enterprise, as it makes IT operations more efficient. Some like its green qualities as virtualization saves on energy consumption, while others appreciate the storage capacity, as well as the data recovery solutions for if disaster strikes.

However, the virtual environment is invisible, and with that come more challenges in making sure it runs smoothly. The cloud might be simple to setup, but it becomes more complex over time. In addition, the more machines and data involved, the more difficult it can be to monitor for space, CPU spikes, network security and other indicators.

“If there is a bug or a discrepancy, I need to know that there’s a problem before my customer does. And though that is the biggest challenge, it’s also a great opportunity,” Russ Caldwell, CTO, Emcien Corporation said.

One of those challenges is making sure storage in the virtualized environment is adequate. “We focus on storage and database environments that scale as the customers grow,” said Caldwell. “Determining how fast customers grow and change is the biggest factor for determining the adequate storage size.”

Companies looking at virtualization solutions need storage solutions that are flexible so they can add or remove storage, as needed. Even though it may have been the right size in the beginning of a project, things change, and a flexible virtualization tool can give that peace of mind when things change. For example, when we’re working with slow-moving manufacturing data, we can determine the adequate storage size easier than when we’re working with hundreds of millions of bank nodes, where the growth is much more dramatic.

The key, according to John Ross with virtual solution company Phantom Business Development at Net Optics, is to truly assess the performance of the servers and the requirements of the virtual machines. This requires monitoring to be in place for the life of the systems to predict utilization and to modify placement based on performance. “When this is not accounted for, it can appear as though there is high CPU utilization on the hosts as well as the VMs,” said Ross, “With the use of protocols such as NFS and ISCSI, it can put quite a load on the network.”

Companies moving to the cloud also have to change how they think about networking. “It can be hard to understand how network connection works when there aren’t wires to simply plug it into a box, but instead virtual, invisible connections that need to be managed through APIs or online interfaces,” said Caldwell. One of the challenges for a company with multiple clients is keeping client data separate from one another. Grouping machines together and isolating them in their own network is the best approach in tackling this challenge. Using excellent monitoring tools smartly can ensure that the network is as reliable as possible.

“Network connectivity comes down to whether the network connection is a single point of failure: If your virtualization solution is off-site, it’s only as good as the quality of the Internet connection between you and your provider,” said William L. Horvath with DoX Systems. If you have a single connection between you and the Internet, that’s one problem. (You can reduce the risks by contracting with two or more ISPs and getting routers that support trunking.) Likewise, if your virtualization provider’s facility is in a single geographical location (say, Manhatten) that loses functionality for an extended period of time due to some natural disaster, you’re hosed. Our Chamber of Commerce lost access to a cloud-based service not too long ago because someone in the data center, which wasn’t owned by the service provider, forgot to disable the fire suppression system during emergency testing, which unexpectedly destroyed most of the hard drives in the servers.

To avoid the challenges involved in virtualization, Ross provided the following tips:

1. Plan on virtualizing everything — not just the servers but the network, the storage, the security … everything!

2. Standardize everything, from the operating systems on upwards through middleware and applications. The more uniformity exists within configurations, the easier it will be to scale and move these workloads optimally around the environment.

3. Ensure network capabilities are met. This will dynamically change and collapse. There will be huge flow changes as utilization and cloud are adopted.

4. Implement resource monitoring. Existing legacy tools will not provide the data or detail needed.

5. Implement a decommissioning process. Ross repeatedly finds several unused machines running. In a virtual environment, this can become a major issue, consuming resources and driving up costs.

6. Plan for backup and disaster recovery. This will drastically change in virtualization and must be addressed.

7. Train your team based on what the management will look like, not on the migration.

The cloud solves certain problems really well and it allows for SMBs to have the flexible infrastructures that they require — without a lot of capital or hardware or payroll costs. Using the cloud wisely with the right tools, companies can get a leg ahead.

Advertisements

Drastic Measures Not Needed with DRaaS

mike-gault“We are seeing 100-year hurricane cycles arrive every two years.”

Mike Gault

Perhaps the only thing worse than a disaster happening is seeing it coming and knowing nothing can be done to stop it. Businesses along the northeastern seaboard had several days of warning before Hurricane Sandy struck, certainly not enough time to implement a disaster recovery plan from scratch. Even more painful is the understanding that some disaster recovery plans would not be enough; physical backup systems in separate geographical areas may have still suffered the same losses as the home site due to the size of the storm.

Most disasters come with no warning at all. Explosions, power outages, and simple equipment failure can cause the same damage. Operations are down, customers suffer, and revenues tank. Once business recovers the harder work of wooing back customers and convincing new ones about the company’s reliability begins.

Simply doubling up infrastructure and creating physical backups is expensive and time-consuming leading to systems that function inadequately when put into use. Cost cutting means doing without applications and information essential to performance. Lack of testing and differences in tools lead to inefficient work practices during the recovery.

Move into the Cloud

Cloud computing and virtual services eliminate a majority of these concerns. Disaster Recovery as a Service, or DRaaS, is a resource-efficient method of allowing business to continue with little to no interruption. Because everything resides in the cloud, no duplicate infrastructure is needed, testing and upgrades are assured, and no applications or information need be out of commission.

DRaaS is a natural extension of the cloud computing phenomenon. Service providers have hardened their security and created tiered services that fit any budget. Companies are embracing cloud computing for a variety of purposes. The flexibility of such services is a huge driver to adoption since only the services needed are active. The rest can be brought online as desired or shut down during idle time.

IT overhead and infrastructure reductions create cash to fuel growth. Cloud services are the perfect vehicle for the rapidly expanding mobile worker and consumer groups. By taking the time upfront to plan and consider operational requirements, disaster recovery can be the key to successful business recovery.

Service Level Agreement Considerations

The Service Level Agreement (SLA) spells out exactly what will and will not be provided with any cloud service. It is crucial to understand the SLA governing disaster recovery, because a disaster is not the time to discover shortcomings in coverage. Performance and productivity need not suffer if due diligence is taken to make a realistic determination of business continuity needs. Planning wisely also keeps SLA costs to a minimum.

Consider these questions:

  • What applications must be included?
  • What operations are essential for service?
  • What information must be easy to access during this time?
  • How often are testing and upgrades performed?
  • What guarantee of data integrity is offered?

A good service provider will have the experience to help answer these and other questions. They should have an excellent understanding of the extent of disaster recovery needed in a variety of industries. Some providers may even specialize in certain verticals, deepening their ability to determine needs and provide suggestions.

DRaaS Benefits Tower Over Risk

If nothing else, Hurricane Sandy brought home the absolute worst that could happen. Fire, flood, and power failures on such a massive scale are unprecedented but not impossible. Disaster recovery is an essential part of business continuity that must not be put off.  The cost of loss far outweighs the cost of DRaaS because, even if such events are rare, all it takes is once. New York Governor Andrew Cuomo said we are seeing 100-year hurricane cycles arrive every two years.

With the knowledge that DRaaS, like all cloud services, is a cost-effective way to relieve the worry of business interruptions, large or small, business owners can put a line through this item on the to-do list. With guarantees of integrity and continuity, resources and energy can be channeled into growing the business and keeping customers happy.

# # #

Mike Gault is CEO of Guardtime, a developer of digital signatures that algorithmically prove the time, origin and integrity of electronic data.  He started his career conducting research in Japan on the computer simulation of quantum effect transistors. He then spent 10 years doing quantitative financial modeling and trading financial derivatives at Credit Suisse and Barclays Capital. Mike received a Ph.D. in Electronic Engineering from the University of Wales and an MBA from the Kellogg-HKUST Executive MBA Program in Hong Kong. You can reach him at Mike.Gault@guardtime.com or visit www.guardtime.com.