Capacity Management in the Cloud

Unless you've been hiding under a rock for the past 10 years, you will have been seduced by the wonder of 'cloud' computing in some shape or form. For consumers, this may simply be the promise of cheap or free storage in return for brand loyalty (icloud, Dropbox, Google Drive, Office 365 etc), but for IT geeks the pot of gold will have been sold as the much greater rewards of lightning fast time to market, reduced costs, and 'elastic' capacity. 

‍

So by now, most of us will have one or both feet in the cloud, which means that 'Elastic' capacity will take care of everything, and the remaining Capacity Managers will either have their feet up watching the cricket, or be forming an orderly queue at the job centre.

‍

Or maybe not?

‍

The principal of 'Capacity Management' remains as relevant now as it ever was, and in a world where we can bypass lengthy sourcing processes, and just click an icon to add more capacity, it becomes even more relevant that both capacity, and capability are managed to ensure that we only pay for what we need. At the end of the day, 'Cloud' is just hardware in someone else's data centre, so the principle of managing your infrastructure & software licenses efficiently still remains. 

‍

Cloud Service providers enable organisations to add (& subtract) processor power and capacity with ease, but how many of us can honestly put our hands on our hearts and say that we're confident we don't have any wasted resource in our environments? Furthermore, do we really believe that when our developers and testers are given the power to size their own environments, they will not choose a Porsche over a Skoda? (And yes I do know that Skodas are pretty good these days!). 

‍

One of the key virtues of virtual machines is that they can be configured to only be available during working hours, allowing us to only pay for what we consume. Without appropriate monitoring and controls however, our developers and testers may not only be using the Porsche for the school run, but may also be leaving it in the garage with the engine running all weekend! 

‍

It's not all bad though, for most production environments, cloud offers the ability to automate the sizing of resources based on rules enabling automatic expansion & contraction based on load. When configured correctly, this solves the issue of dramatic changes in load faced by web sites selling tickets for the latest boy band, or Telcos struggling with a 1,000% increase in SMS on New Years eve. However, we still need to ensure that when Services are approaching end of life, Infra, Storage & Software licenses are reduced at the appropriate point, and eventually decommissioned. In large organisations there is still plenty of low hanging fruit in this area.

‍

So the gardening leave is over, Capacity Managers are back in business, and we're dusting off the ITIL manuals again. The IT landscape has changed, but one thing remains the same, IT will still be the last to know when the business plans a promotion, and regardless of how much 'elasticity' you have in your capacity, be prepared for a meltdown!

‍