Going hybrid the other way – with OpenStack
A friend of mine, Chris, a former CIO, said that if he started a company today, he would never build and operate his own IT in favor of using public cloud services. With a clean sheet of paper, why get into the business of architecting and managing software, servers, networks, and storage? At face value, this sentiment made sense to me. Given the abundance of public providers available, it seems you can get all that you need to run a business on a pay-as-you-go basis. Why then am I hearing about companies choosing to migrate services out of a public provider and into their own private cloud?
I guess there is the shocker moment when the once-small-but-now-huge monthly bill finally comes to your attention (There goes our OPEX budget!). Or perhaps it’s a latency and reliability thing. Or security. For various reasons it seems, you can reach a point when it starts to make sense to look at alternatives and at bringing things back into your own domain of control. It’s not that the public cloud is completely out. It’s just that many are going hybrid the opposite way than most people think about it. I suppose hybrid is hybrid – no matter how you get there.
But what about that ease-of-provisioning – a few clicks and I’m good? Do I give that up? Moving off the cloud to a private data center – it’s like a bearded Spock in the Mirror Mirror episode of Star Trek – this isn’t the way it’s supposed to be – is it?
This is where cloud building tools like OpenStack aim to be your friend when you’re ready to begin rolling your own cloud. I see a lot of interest and momentum in the technology in both the end user and IT vendor communities. OpenStack – the open-source cloud-building toolkit – helps cloud builders deliver infrastructure as a service (IaaS), both public and private. This week is the OpenStack Summit (Hedvig will be there). One of the sessions is focused on this topic of migrating to a private cloud OpenStack deployment from a public provider. If you’re at the show – check it out.
In a similar-but-different approach, we are working with a company that wants to take their storage out of the cloud but want to keep using cloud-based compute. Since the Hedvig Hedvig Distributed Storage Platform supports both private and public cloud deployments and interfaces like S3, Swift, Cinder, and REST APIs, a hybrid storage and compute deployment is a real use case for us. To make it feasible, you need to host your private storage cluster nearby to ensure low latency – but since it is not uncommon to have co-lo sites in the same datacenter as a cloud service this can be entirely possible.
At some point – whether with OpenStack, Hedvig, or similar type of infrastructure technology provider, I expect this kind of capability – to move up to or down from (migration out of and into) clouds, private and public – to become table stakes. Getting to real cloud with automation – versus being simply virtualized with some workload mobility – depends on it.
So I say to my friend Chris – yes, let’s start in the public cloud – but don’t be surprised if your CIO-hat is back on in the near future and you are guiding the orchestration of your own private cloud infrastructure with OpenStack (and Hedvig!).