Red Hat's John Mark Walker: The Open Cloud Needs Open Storage

John Mark Walker: The funny thing is that storage, like many things, is morphing and merging with a host of other technologies. If you look at the industry trends that make up the open hybrid cloud, elastic, automated storage is quickly being recognized as an intrinsic part of any cloud architecture. The problem is, as with many technologies, those responsible for deploying these open hybrid clouds are discovering that the traditional, proprietary approaches don’t scale. That in order to have an open hybrid cloud, they need an open hybrid storage platform that’s agile enough to support their data center’s rapid growth.

 

Historically, compute was the sexy part of cloud computing, while storage was the backwater that nobody really cared about. And then storage requirements started to exceed the ability of operators to keep pace, and suddenly storage was recognized as something that needed to be just as elastic as everything else. The problem was, storage means data, and nobody can afford to lose data. If you lose compute resources, spinning those up is easy, and you can continue wherever you left off when the old compute node died. But that implies that there’s a storage layer underneath allowing you to preserve the state of all of your virtualized machines and apps. If that piece goes away, you better have some mechanisms in place that allow you to quickly recover. So, in our brave new, open hybrid cloud world, you must have an open hybrid storage counterpart that’s baked into your data center architecture. 

Walker: The Gluster Community is all about creating software that allows you to achieve total data center victory. That is, invest in the architecture that gives you the open hybrid storage layer that we provide, and you’ll have be able to produce a system with extreme data portability and mobility. We’re producing a unified storage backend that gives admins maximum data flexibility – nobody else can say that right now. One of our central guiding principles is that your access shouldn’t be limited by the protocols that your app speaks. If you’re using NFS, you should be able to access the same data as if you’re using Swift, S3, or some other object storage protocol. Your data, your way – your vendor shouldn’t be able to dictate to operators how to access data.

Walker: We are currently in the process of transforming the Gluster Community into a vibrant open source ecosystem. I’m afraid I can’t share more than that at the moment, but expect some big changes coming up very soon. This is, by far, the most exciting.

Walker: Lots 🙂 There are some substantial changes coming up in GlusterFS 3.4. Up until now, we have been almost exclusively focused on scale-out NAS: files and folders distributed over a wide network. One of the hottest areas of growth right now, thanks to cloud proliferation, is the ability to manage virtual machines running applications in the cloud. The ability to replicate these machines for failover in the cloud has been something that we weren’t able to do at scale. However, with GlusterFS 3.4, thanks to some major contributions from our friends at IBM’s Linux Technology Center, we have built tight integration with QEMU/KVM, allowing operators to deploy VMs on GlusterFS at a large scale and with better performance. We are very excited about this – both because it’s a great feature to have, but also because it marks a very important milestone for us: This is the first major feature contributed from a non-core GlusterFS engineer – ever. This shows how much we’ve grown from your typical open core model to a real open source model.

Walker: You’ll be hearing about Total Data Center Victory, obviously 😉  Really, this talk is about two primary themes, the first being how much of our data center world has been led by open source developers. I’ll give you a hint: pretty much all of it. We have arrived at the point where industry innovation is not led by proprietary software companies,  but by open source communities featuring informal, de facto partnerships between innovative software vendors, like Red Hat, and innovative operators and service providers, like Netflix, Twitter, Facebook, etc. This is the way forward – we have seen the last proprietary innovator in the data center.

 

We always wondered what total victory for open source would look like – we see that right now. Now the question becomes, what now? Where will open source lead us? When you look at the pace of innovation in ecosystems like those centered around OpenStack and Gluster, we’re heading towards a landscape of total open hybrid automation – the open hybrid cloud converging with open hybrid storage. When I say “Total Data Center Victory” it’s a double entendre – Open Source developers have achieved Total Data Center Victory by following their muse. And today’s Linux pro can achieve Total Data Center Victory by using these trends to their advantage. Take advantage of the converging worlds of data and compute. Take advantage of open hybrid automation and be a part of the communities that are driving innovation.