The impact of virtualisation on storage and backup

Andrew McCreath, virtualisation practice lead at GlassHouse Technologies in the U.K., explains the impact of virtualised server environments on storage and backup.

Q. What effects do virtualised environments have on backup and storage?

A. The primary effects on backup and storage are around utilisation. With virtual machines you obviously increase your utilisation of the storage layer by not overallocating physical disk, and on the backups you improve efficiencies. The backup environment can also change in many ways from your current methodology. Whereas today you may have a backup machine with a virtual client installed, you may move to a host-based or virtual consolidated backup. Each of these will deliver higher cost and better utilisation across your backup and storage estate.

A virtualised infrastructure has allowed organisations with siloed technology to move to a shared infrastructure model. Whereas previously an organisation with a finance, marketing, HR or sales division may have had their own projects funding isolated silos of SANs [storage-area networks], they're now moving to a consolidated, shared infrastructure. It's worth mentioning that shared storage needs to be capable of delivering the I/O on the virtual estate as well. A lot of people make the assumption that their current technology stack will just migrate to the virtual environment. But a lot of companies will need to address their storage needs and upgrade, enhance or work around the existing technology stack they have.

Q. What are the key tasks to optimise storage and backup in a virtualised environment?

A. The key tasks for backup are identifying the business requirements, the data required to be backed up on an ongoing basis. Historically, a lot of organisations when deploying virtualised environments haven't looked at what backup changes need to happen. They may have had a day one policy that's just evolved and grown.

The general practice is to deploy some good housekeeping measures: identify legacy data and clean it up before you migrate. This will lead to a cleaner, more efficient storage and backup environment. Some key tasks around the storage will be identifying the configuration of the disks to ensure performance, ensuring your zoning set is correct across all hosts, and ensuring all the technology stack and connectors are on the hardware compatibility list and supported infrastructure from the vendors. These tasks can be overlooked, usually when organisations have their own reference technical architecture and they just carry on using that. Organisations need to review and adopt new technology and best practices in line with virtualization, and they will need to ensure they have a good understanding of it. This may involve recruiting key resources or using consultancy skills to fill the gap. Key tasks in backup will be ensuring your policies are accurate, that they are optimised in the current world, that housekeeping has been applied to the data set that you're backing up and ensuring you're not duplicating virtual machine backups. It's very easy to take snapshots of virtual machines and keep snapping them, but you'll need to have processes to clean those up on a regular basis.

Q. How do virtualised environments affect capacity management?

A. Effects on capacity management primarily affect storage and its utilisation. With many federated physical server estates you'll find utilisation is around the 40% to 60% mark. This is because you buy a server and typically it has a 140 GB or 73 GB hard drive installed, and yet the operating system or data may only come in at a half or third of that. When you move to a shared storage infrastructure, like a SAN or NAS [network-attached storage], you have the opportunity to streamline that.

Many projects come into an organisation with a focus on a forecast data set of maybe a three or five year period, and that is allocated to the budget up front so capacity management will then procure and deploy the required capacity for that period, leading to overallocation. With virtual machines, they will only actually use the disk requirements in place today, so if an operating system has 8 GB it will use 8 GB rather than the fully allocated 12 GB. This leads to better utilization, which leads to better ROI on the storage and better value through the virtualisation programme.