It’s that time of year where we all come up with a few ideas for New Year’s resolutions.
For the datacentre manager, here are 10 New Year's resolutions that may come in use for putting in place a technology platform for 2014.
Before we get on to the datacentre facility itself, it is important to start with the IT equipment. Has this been optimised for purpose?
Do you know how many applications you are running? How many different instances of these applications there are? How many servers you have? How many operating systems, and at what level are they all patched? How much storage you have? How many network ports are not being used? Carry out a full audit of your IT estate – this will create a great starting point for everything else.
If you find that you have different versions of software running, bring them all to the same release and patch levels. This makes management so much easier: multiple instances of the same application with different versions push management costs up to inexcusable levels.
From the audit, you may find that you have multiple applications doing essentially the same thing. Cull those that are superfluous and bring down the number of applications to as few as possible. This will help in to centralise your data so it it can be analysed more effectively – multiple applications give rise to multiple silos of data that are difficult to deal with.
Look at your storage systems and clean them all up. You could be amazed at the wastage that you could find in how many of the same, or very similar, files are being stored. A good de-duplication approach can lower storage needs by 80% - imagine how that could impact storage buying decisions for the next year or so. Make sure that the data you have is “clean” – use external sources to check that customer data is correct. Use master data management to ensure that new data is created through a single source of the truth.
For some, this may be old news, but Quocirca finds that many organisations have not progressed far along the virtualisation route. VMware, Microsoft, Citrix and other hypervisors are all now proven, and organisations should be looking at not only virtualising servers, but also storage and networks to make for a more flexible and available platform.
Building your own rack-mount systems is no longer the best way to create a flexible and optimised IT platform. Modular systems, such as VCE VBlocks, Cisco UCS, IBM PureFlex systems and other pre-configured approaches provide far better performance and faster time to implementation.
Cloud may still be at its early stages, but it should be on every datacentre manager’s list for 2014. The capability to flex resources to meet the different needs of multiple workloads means cloud is a major way of optimising all available IT resources. However, it can only be done effectively if all the above steps have been carried out first – do not just throw money at putting cloud in place if you have no idea what workloads you are running, or have not carried out data cleansing or started using virtualisation. Jumping straight to cloud could just mean throwing a lot of money down a very large drain – and having to roll back to a less optimised platform when it all goes wrong.
The big question that should be asked by any datacentre manager – even though this may be the painful one for them. Just how future-proof can an owned facility be these days? With the current rate of change at the platform level, as virtualisation, cloud and converged systems all make an impact on how a datacentre needs to be configured, it may be best to look outside of the owned facility.
At the very least, a co-location facility may offer better capabilities in managing an increasing equipment densities with the consequent needs for more power and cooling. For some, it may well be the abdication of IT as a platform completely with a move to public cloud-based systems through the use of software as a service (SaaS). However, for most, it is likely to end up as a hybrid mix of co-location and public cloud, with many still having a rump of an owned datacentre for certain workloads.
If you do choose to retain a datacentre facility, then optimising energy utilisation should be close to the top of your priority list. Increasing equipment densities require far more power to be provided to racks and converged systems – ensure your facility can manage this, and manage it in a high-availability manner, with multiple redundant power distribution systems. The amount of energy that is used by cooling systems in private datacentre facilities is still far too high – the latest research shows that this could be more than 2.5W for every Watt that is used by the IT equipment.
Look toward the use of “free” cooling systems, such as adiabatic, Kyoto wheel or other systems, and use hot and cold aisles and/or contained racks and rows wherever possible. It should be possible to bring pretty much any facility down to a position where less than 1W of power is used by the facility for every Watt that is used by the IT equipment – and your aim should be to make it 0.5W or less. The energy savings will be worth it – and the payback on investment rapid.
Forget that people will need to go into the datacentre. Make it as self-sufficient as possible; run it hot; manage it from outside. Put in place the right tools to manage the total environment from a distance. Use datacentre infrastructure management (DCIM) tools from the likes of Nlyte, CA, Siemens and Emerson Network Power to give a complete picture of what is happening. Keep people out of the datacentre as much as you possibly can – it will be better for it.
IT is the lifeblood of today’s organisation. As a datacentre manager, you have to be in a position where what you put in place and manage is best fit for the organisation’s purpose. Using the above as a template for your 2014 resolutions could put you in a better place running the organisation: it could save your job.