Realistic planning

I mentioned previously that virtualisation is not the legendary silver bullet.

It is not a panacea for all the problems that the IT staff, the company infrastructure or the business itself has to deal with in a challenging economic environment.

In fact, incorrect planning of a virtualised solution can either bring limited or even negative returns on the investment made.

The problems typically fall into five main areas: replacing hardware sprawl with virtual sprawl; power usage issues; security and authorised access obstacles; integration with current systems and processes; and overall systems management.

There are few production environments that do not suffer from sprawl in some way. It builds up as new systems are added and is also inherited as essential legacy systems are carried forward into new network architectures. Without adequate planning, it is easy to carry this patchwork model onwards and end up with a virtualised solution that too closely maps the existing network. This is less than optimal.

Before extensive virtualisation is deployed, there has to be a realistic assessment of what applications and processes are critical, including which of those will most benefit from being virtualised. This requires more than just understanding the technology. It requires an overall view of the business, its needs and the needs of the people involved. Only then can the virtualisation programme be mapped out, which will deliver the best balance of amount invested and results achieved.

A company needs to know what its current resources are before it plans to implement virtualisation.

Andrea Lodolo is technical operations manager of CA Southern Africa.

It is often assumed that virtualisation will save on power usage. That is not necessarily the case. Servers that were under-utilised can be reduced in number, but the remaining ones will be drawing more power, including the roughly equal amount needed for cooling systems.

To achieve power savings there has to be a delicate balance. This cannot be achieved without accurate reporting and management of the power usage, which means reporting technology must be an integral part of the virtualisation.

Security is of paramount concern, but another delicate balance has to be struck between really solid security and ease of access. This is even more important in the modern business, where remote workers might log in from any location or device.

Whatever virtualised solution is put in place, the same rigorous security must be applied as for the previous network solution. This must also have the flexibility to allow single sign-on that is not limited to specific machines and locations. This might require more planning and technology than the basic directory service, and it might involve necessary changes to provisioning and de-provisioning of passwords or other authentication.

In the area of integration, the first question is whether existing infrastructure is fully enabled to support virtualisation; whether it is internal or outsourced. This deserves careful analysis, which includes looking at the hardware, the connectivity and the software. Necessarily, it also includes looking at the critical business processes and the applications they depend on. Can these be virtualised? Can they operate seamlessly with components the company plans to virtualise?

The answers can only be reached if there is a complete audit of network resources and applications, weighed against the business needs.

On the systems management side, the challenges of a virtual solution are the same as those of any other network. Depending on the planning done at the outset, there might be advantages because a virtual network can be easier to automate for reporting and failure notifications. The downside is that the virtual components might need additional systems that have to be integrated with existing management software.

To resolve these issues, it is important to have an end-to-end systems management solution that handles the whole network. This avoids a situation where disparate systems have to be patched together to provide comprehensive reporting and management functions.

There are two key points that emerge from all the challenges listed above.

Firstly, a company needs to know what its current resources are before it plans to implement virtualisation.

Secondly, it needs to plan the deployment in a way that takes into account the overall network solution and all the business needs.

Virtualisation will not achieve the desired results if there is only a vague or general idea of what hardware and software will interface and integrate with it. It can greatly rationalise business processes, but only if there is a clear understanding of the point of origin.

Virtualisation will bring savings and improvements, but to get the best value for the business and the fastest return on investment, it has to take into account the realities of the production environment, what the business as a whole really needs, and not least, the needs of the users of the system.

None of this can be achieved without IT management solutions in place that deliver accurate information and which conform to established best practices. These have to be used before, during and after the deployment of a virtualised solution to ensure that the virtual solution achieves its goals and doesn't just carry over some of the same issues that existed before it was rolled out.