…or is now a really cool time to be working in Infrastructure? We have “Cloud”, “SDN, “NFV”, OpenStack (and its myriad of components) and Docker to name a few of the plethora of generally meaningless words and acronyms. You know when a technology is really hot – it’s when it’s the random item shoehorned into practically every vendor presentation (or LinkedIn Post 😉 ) to show they are current. Yet for all the trends and their path along the Gartner Hype Cycle it feels that things are really coming together to mean real changes are happening.
As someone who works in Infrastructure and who has to produce those wonderful pin the tail on the donkey things known as roadmaps I feel genuinely excited about the transformation in the air. Looking ahead we will go from monolithic teams and budgets to Service brokers across multiple cloud providers. We will transform from handy fall guy for any delay or incident to genuine enablers. If everything goes according to plan we’ll become as invisible as any decent utility company with no need ever to talk to us (probably for the best!).
There are things we can do today which enable servers to start to become more portable – able to float between datacentres and even up in to the public cloud providers. These have massive impacts on previously gospel concepts such as DR. DR was easy, if its important have two of it. If you are a really big company have at least 3 – great for hardware vendors! Now though you can have DRaaS (DR as a Service) where you can “reserve” capacity in a virtual private cloud and only inflate it when you have a Disaster – compared to kitting out another DC and managing it result!
Inter-connect providers allow us to start consuming public clouds without being penalised by internet type speeds. Network virtualisation means we can span or duplicate our networks across traditional boundaries allowing workloads to wander around freely from a network point of view (yes there’s storage etc. one thing at a time seriously!). “Burst” capability where we create workloads in off-premise 3rd party clouds to serve short-term demand mean we only need to buy and maintain sufficient hardware for 90% of our demand, and we can rent the peak 10% as and when it happens. It also takes the guessing game out of trying to serve that extra peak – the sales target and infra might be for 10% growth the reality could be 20% growth – without burst you’re in trouble, with burst you just spin up more servers (and hope the backend systems can cope I know but work with me here!).
Application teams need to do their part to assist and methodologies like the 12 factor app (http://12factor.net/) combined with containerisation be it Docker or CloudFoundry will assist to make workloads truly portable. Once we’ve cracked the portable workload challenge if we couple that with automation technologies we can move workloads between cloud providers following the best price without impact to the Service – great news for Corporate Purchasing teams!
As our automation tools drive us to define more and more in software we can start building out a picture in software of how an application hangs together. From there we can start building dependency trees so in a Disaster we don’t just use continuous delivery pipelines to build new instances but build them in the correct order – rather than the nightmare process of restarting things until they’ve all come up in the right sequence.
As brokers we can leverage these toolsets to enable the developers to just develop, not worry about infrastructure, about availability zones or DR. Just to do what they Love which is (hopefully) code!
Its not all about the public cloud though if you’re working in a large Enterprise with a couple of DCs and you’re sat on a large amount of data which you’re not comfortable putting in the cloud you’ll probably hang on to those DCs (having said that a major US bank this week announced it was all going Public Cloud). They may get smaller as other workloads move out, but fundamentally you still have a DC and if you have a DC to make it cost effective you probably ought to run as much as you possibly can out of it. OpenStack is getting ever more mature to the point that it provides real opportunities to drive down the cost of your private cloud and yet still offers you the option on sticking on your preferred hypervisor.
As you have probably guessed I can could ramble on for much longer but you get the gist – there are a lot of possibilities out there right now and it’s a great opportunity to drive down costs while truly step-changing capabilities!
Please note that this post was orginally posted on LinkedIn
No Comments