The top two common problems when it comes to implementing virtualisation involve storage, explains Evan Powell, CEO of Nexenta Systems.
First, customers do not include the implications to their Storage needs when implementing virtualisation. Two, customers very often try to force fit existing storage solutions to meet the storage needs for virtualisation. Ex-ante virtualisation is considered as a silo project with the objective of maximising compute resources. But there are downstream implications that frequently result in cost overruns.
These issues can be prevented by creating a framework that considers storage requirements to meet the new virtualised environment. Run through a design that considers the end to end details of your infrastructure including, compute, network, storage and switches. This will help you to identify the needs of the entire project upfront.
Customers very often tend to use their existing Storage technologies and force it to accommodate their virtualisation needs. Simply put, the traditional storage appliances that were built to handle the new requirements of virtualisation. The most common mistake is to assume that virtualsation can be achieved without changes to the entire infrastructure stack.
Consider alternatives to traditional storage that are purpose built to manage virtualised environments. OpenStorage technologies help customers to bring the best of hardware and software infrastructures to meet the new requirements. Simplify the complexity by de-coupling the hardware from the infrastructure stack.
Regarding a strategy, the first step is to ensure the local network is capable of handling storage virtualisation. Intermittent network performance and connectivity issues can quickly reduce its effectiveness and have an impact on storage performance, says COO of Oncore IT, David Ebsworth.
What is essential is guaranteeing that the hardware deployed offers an equivalent or better IOPS (Input/Output Operations Per Second) rating as localised storage. Virtualisation offers substantial cost benefits, but witnessing a reduction in disk performance will negate the advantage of virtualisation in the first place.
Put simply, the disk technology has to be fast enough to cope with demand. Alongside high performance, we also need to consider expandability. Reaching critical mass can halt business operations, so when virtualising a storage environment possessing a strategy for data expansion is crucial for staying ahead of potential complications.
This links well to disaster recovery. Not only should expandability be a concern, but replication to secure failover sites needs to be considered. The storage systems, as well as the data, need to be replicated for the business’ continuity. Ignoring disaster recovery is a dangerous choice, especially when an entire enterprise’s storage has been virtualised.
Finally the cloud; do not blindly trust unproven providers who hide behind its veil. Pursue a partnership with a provider that can concretely demonstrate the upkeep of its service level agreements. Transparency is everything and this translates to return on investment. Do not pursue the cheapest solution as it will only result in expansion and performance limitations that reverse the benefits of virtualisation.
The goal of virtualisation is to have the capability to move and scale
applications instantly without worrying about the physical layout of the
datacentre - in effect, to centralise the management of servers in the same
way SANs have centralised the management of storage, explains Jamon Bowen, Director of Sales Engineering at Texas Memory Systems.
Moving the computing part of an application is easy, it is access to the data that the
application depends on that is the difficult component. Enabling this
flexibility requires a high performance network and high performance storage
on the back end. In some cases, however, it just does not make sense to
build an extremely high performance network and storage backbone if most of
the applications do not have higher performance as their ultimate goal.
What we are seeing is that IT teams and services are fragmenting into two
camps: the cost centre and the profit centre applications. With the cost
centres, all of the optimisation is targeted to reduce overhead and
management. These applications are all being virtualised now and will
eventually end up in the cloud. For the profit centre application,
performance is the priority, and we are seeing a shift towards rolling back
the levels of virtualisation and more focus on hardware layout and
optimisation. There is even a resurgence in the direct attached storage
model with solid state disks (SSD) for these applications. These
applications are the core of what an organisation does in the market, where
higher performance = more profit.'
Tags: Storage Virtualization