CIO and business requirements for a flexible, responsive, efficient computing infrastructure that demonstrates significant ROI has already been clearly demonstrated by the business case for virtualisation. However, what do CIOs need to consider before virtualising their environment?
Chris Norton, regional manager for VMware SA, provides seven steps CIOs need to consider when embarking on their virtualisation initiative. Demonstrated cases of virtualisation around the globe and locally have shown that planning is a key element to virtualisation success. So, how do companies get it right?
1. Properly define the business case
There are many different reasons organisations decide to go the virtualisation route. For example you may need to get server sprawl under control or reduce your data centre costs significantly. You may find that security and virus control is a huge issue and minimising downtime is a key driver for your move to virtualisation.
Whether it is one of these, a combination, or something else altogether driving your decision, ensure that you are very clear on these issues and document them. Clearly define the processes of the project. Know upfront what cost savings you expect through virtualisation. What third party application support policies are in place and how will this affect the project?
Once you have built the business case you can set about getting agreement from others within the business. Getting buy-in for virtualisation has its challenges. Business units are used to ‘owning’ their physical servers and keeping their applications and data in silos away from others. The practice and habits that come with decentralisation will require changing and as such you will need executive commitment and support from the CFO. Whatever the objections to virtualisation, know they will be there. Having a documented business case for the introduction of virtualisation and the flexibility it brings into your environment will help its adoption.
2. Phased implementation is crucial
Taking a phased approach is critical when implementing virtualisation. By splitting your roll-out in phases for example via department or application segment or region, you are able to move carefully through each phase ensuring problems are attended to and developing the virtualisation skills of your teams. Your first phase which is usually the pilot phase, will provide you with a success case for your other migrations. In addition, by starting with a smaller amount of servers first you will be able to adapt your processes and strategy upfront if required.
3. Performance and consolidation ratios must be considered
The number of virtual machines that can be run at target performance levels on a given physical machine is referred to as the consolidation ratio. For example: 10 virtual servers on one physical server = 10:1. Understanding your consolidation ratios is important for when you design and deploy your virtual infrastructure.
If you have a high consolidation ratio say of 15:1 or higher, the impact of a host going down is large and this needs to be considered when deploying disaster recovery elements of virtualisation. Aim for a 65% utilisation of your server hosts as you will need some flexibility to move virtual machines around in the event of planned or unplanned downtime.
4. Peruse your current environment carefully
It stands to reason that if your servers, network or storage is badly configured in the physical world they will be badly configured in the virtual world. If your applications are outdated or unsupported by the business, why virtualise them? Review your application and computing portfolios before virtualising your environment to reduce costs for virtualisation as well as overall support costs to the business.
5. Pursue an evaluation of your current physical environment
Doing an evaluation of your existing physical environment should be completed before going ahead with virtualisation. Evaluate your server utilisation stats, access to storage such as SANs and NAS, network topology and requirements, disaster recovery implications, energy and cooling requirements as well as rack and space availability. Without this stage the design of the virtual infrastructure may not accurately reflect your actual requirements.
6. Plan for shared storage
You need to have a well-defined storage provisioning and management process because to take advantage of the flexibility of a virtual infrastructure, shared storage (SAN or NAS) must be in place. Ensure that you design redundancy in all levels of your shared storage and make sure vendor best practices are adhered to in the management of that system.
When you have a virtual infrastructure in place you introduce network virtualisation to your environment. As most 1 GB NICs are highly under utilised, allocating numerous virtual machines with a single pair on 1 GB NICs allows you to better use the hardware. However, they may be on different networks, VLANs or need different security contexts.
7. People skills need to be considered
It is important to have the basic levels of virtualisation skills in-house to execute and resolve baseline issues. Review the skills sets that you currently have and ensure you enhance their skills sets accordingly. You will need the following skills to design and deploy a VMware Virtual Infrastructure 3 enterprise deployment:
* VMware Certified Professional (VCP) Certification.
* Basic to intermediate Linux skills.
* Basic to intermediate backup and recovery skills.
* Shared storage (SAN/NAS) knowledge.
* Intermediate networking knowledge.
* Advanced server and OS skills.