Geographically mobile compute is a possibility as virtualisation technologies broaden, says Mark Hyde, Fujitsu NZ solution architect.
I picked up a t-shirt some months ago at a conference emblazoned with the words, ‘Virtualise Fearlessly’. I didn’t pay it much attention at the time, but it is something that has come to mind from time to time.
Those of us wading regularly in virtualisation probably wouldn’t have agreed. Another nugget I picked up was, ‘If you can get IT a bit wrong, then virtualisation would help you get it a lot wrong’. The extra layers of virtualisation do help to compound simple errors into serious problems. So perhaps the shirt should have read ‘Virtualise Carefully’.
Recent months have finally promised to put us in total control of our virtualisation environments with a swath of monitoring tools and both networking and storage joining the party.
The nirvana of running all physical resources at around 80% of capacity, with the operational visibility we need, without deploying a raft of expensive third party tools, is now within sight.
The software defined datacentre (SDDC) is now a feasible reality for those companies who wish to grasp the opportunity.
At one recent virtualisation conference, VMworld 2013, networking was getting the same focus and treatment as x86 server architecture has had in the past.
The aim is to deliver the entire networking and security model (L2-L7) in software, decoupled from networking hardware.
This offers exciting and powerful new operational models for networking that break through physical barriers and enable data centre operators to deliver better speed, agility and elasticity to networking.
This is enabled through logical networking, firewalls, load balancers and VPNs all existing within the SDDC and more importantly connected to the physical networking topology through hardware partners.
This offers a truly portable and rapid path to provisioning networking solutions for the modern datacentre.
One customer is looking at embracing this simplification of their networking into the software stack to reduce both the capital cost of a potentially complex network refresh and also reduce the reliance on expensive operational resources for simple tasks that can be offloaded to the software.
In addition, with the future focus on VM-centric storage rather than the traditional volume based approach, it will allow us to build storage and replication policies around compute resources rather than the storage containers which have added additional complexity in the past.
Take all this together with simplified, configurable and intelligent monitoring and management tools and we open a possibility of compute not being tied to a server or a cluster but instead being geographically mobile.
It begs the question; if the datacentre is treated much as the node in the cluster is today, could it be that the reliance on the highly redundant datacentre will be reduced?
There is a lot of change afoot and with broadening of virtualisation technologies and simplification of tools we have put the SDDC within reach of most.
Now, perhaps, we have reached that time where people are comfortable to ‘Virtualise Fearlessly’.