Many companies have been moving to Virtualisation over the last two years since the hype has hit the market, but many of the deployments have been classed as failures by management within the business. This is caused by many businesses taking the decision to virtualise core services, without the correct planning and expectations. In the end, they have major performance issues and degredation

It is true that physical servers are under utilised. According to reports, the majority of servers are used approximately 15% of the time that it is powered on. So how or why are these deployments failing? Well, usually bad resource planning and undefined project definable deliverables and I will try to outline some of the major failure points in this post.

1. Resource Planning

When planning a Virtualisation deployment needs time. In a multi server environment you need to monitor resource usage of your existing physical servers, or have a solid understanding of their needs when planning the architecture and specification within your Virtual Infrastructure. You need to know, which services and physical services can blend well to ensure that a physical node isn’t overloaded and a performance decrease is witnessed.

Key items to monitor include:

• Processor idle time
• Page File use
• Free Memory
• Disk Queue Length
• Bandwidth

2. Additional Network Complexity

Consolidation of your physical environment into a Enterprise/Private Cloud has many business and IT advantages, which include reducing the network complexity. Ironically without the correct training, many IT Administrators and Support Teams are failing to be able to support the the solutions correctly due to a poor understanding of the architecture.

3. Hidden underlying costs

Tied with number one, poor planning of an installation can leave companies needing to add additional servers and complexity into their Virtual networks that weren’t initially planned. With the addition of more servers usually comes the deployment of a Storage Network which adds another costly layer to the deployment.

4. Unrealistic Expectations

In a Virtualised environment, you are sharing server resources. It will be inevitable, that there will become a queue on a specific resource demand at some point and a performance drop will be seen over ths queue length.

You cannot expect to receive the performance of a single server from within a Virtual environment without the correct investment.

5. Not utilizing High Availability

One of the best features that Virtualisation and Cloud computing offers is the ability the fail over ability. The Virtual Machines are not dependent upon hardware, so can be simply configured to move from server to server in the event of failure.

Now as this can be a “Server Consolidation Exercise“, many companies do not use this feature and this can leave them in peril. In our current networks, if one server fails, one service is unavailable. However, if a virtualised node fails, all the servers hosted on their fail, potentially taking down the entire network.

Businesses suffering extended IT outages are statistically more likely to suffer massive financial losses and potentially fold.

Is it all bad?

Definately not, No! Ancar B is an advocate for Virtualisation, and with the correct planning and budget all SME users can take advantage of the benefits whether it be in house or hosted.

Had a bad experience?

If you’re working on a current virtual network and you’re suffering with resource overuse or performance degredation, I suggest you take a look at your performance monitors and see where improvements are needed or bottlenecks occur.

If you’re in a Windows environment, check Microsoft’s Measuring Performance on Hyper-V to help shed some light.

Hopefully, this post will allow you to avoid any issues in depolying a Virtual Network, or just a little insight into the traps that are out there.