I am going to talk a bit about public cloud. One of the main benefits that Public cloud services inherently have, is the ability to utilise the extreme volumes of infrastructure in order to reactively burst and support additional workload, this is more commonly known as Cloud bursting. Cloud bursting is only really made achievable by the economies of scale of public cloud services, with services having coverage across multiple datacentres all geographically located.
When it comes to Public cloud burst I have always been a bit skeptical on embracing the functionality, and this is mainly due to me naturally always being conscious of efficiency within any architectural design and strategy. In ye olden days, Cloud bursting is/was rarely achievable in any other infrastructure architectural model such as private datacentres, primarily due to the requirement to have infrastructure dormant on standby ready to service request (incuring excessive cost) and in order to facilitate a burst in workload would require upfront over provisioning of resource.
A question I ask myself (and readers might), is why does public cloud bursting make this any different? Cloud computing is based upon an operational cost however infrastructure still has to remain dormant (and be paid for indirectly) in order to be able to facilitate both your companies own burst strategy and the burst demand of the 101 other customers that you share services with, so is this really as efficient as older strategies of just provisioning hardware to prepare for a burst?
Lastly putting the practical issues of cloud bursting a side and looking back on the mistakes of the past, when you look at issues such as VM Sprawl and dreaded upfront server over provisioning, this could suggest that cloud bursting will be end up being no different. Bursted capacity will need to shrink down to its original non peak size in order to remain cost effective, so do you think developers are actually considering this in any application design and strategy? Or is cloud burst just the next big problem along with VM sprawl etc that will plague IT?
Lastly there is also the risk this might even get worse with another common old problem, many non IT business units are potentially being sold SaaS/PaaS based cloud solutions that are marketed and pitched to be able to “burst”, and with no careful control education and wisdom being provided to the business on the potential issues that this might end up turning into based upon previous incarnations of problem.
This wasn’t an insightful post you may say, however I just thought i’d provide you with a shade of reality that might just not be something you’ve yet to consider when exploring public cloud.