Choosing the right storage layer for your OpenStack cloud
Mon 27 Jul 2015
Andy Roberts is field architect for SolidFire International. He was SolidFire’s first systems engineer outside of North America, focusing on clarifying the benefits SolidFire provides in OpenStack, CloudStack, and VMware solutions.
Virtualisation has arguably been the biggest shift we’ve seen in IT in the last 10 years. This technology helped simplify IT management by removing the silos between servers, storage and networking teams. It has also allowed applications to be engineered in such a way that don’t need to rely on an underlying hardware platform – they can potentially move in and out of the cloud.
However, organisations now want the ability to deploy application changes in real time – ideally from development to a live environment at the click of a button – which is something that can’t be done with virtualisation on its own. IT environments are increasingly becoming fast paced, causing headaches for the IT department because existing technologies and resources are increasingly strained. As a result, we are now at a transitional period within IT, where the main focus is on increased use of automation.
To address complex automation needs, many people are now turning to OpenStack. According to an OpenStack User Survey, key drivers for adopting OpenStack include the ability to bring innovation to the business: its automation means there is less infrastructure maintenance, so more time can be spent on innovating features and functionality. Its open technology and the lack of vendor lock-in are also really attractive qualities because both can help save on costs by cutting the need for licenses and initial, up-front investment.
As a result, OpenStack is being seen as a major part of next generation architecture; perhaps even the universal OS of the next generation data centre. Indeed, ask anyone who’s worked with OpenStack and they’ll tell you of the huge success and benefits they’ve had from the small deployments they’ve started off with.
But what happens when you want to deploy OpenStack more widely in an enterprise, where reliability and guaranteed performance are key? When you look at large enterprises like eBay, MercadoLibre and PayPal, they’re operating at such a scale that they don’t want arduous management layers between OpenStack and the storage system, or to waste time manually setting performance levels and managing volumes. They need to choose an infrastructure that leaves them free to focus on the core of their business.
Choosing an OpenStack storage layer for your deployment
There are an endless number of architectures available when building your OpenStack infrastructure for the first time. The good news is that companies like RedHat and Mirantis are making it easier to fit the storage and cloud layers together. And these are a good starting point for building your OpenStack cloud.
But if you want your own OpenStack deployment to be more automated, you can only do this by selecting a storage layer that supports this. A few considerations to keep in mind during this process include:
1. Will it scale out linearly as you grow?
2. Is it proven with OpenStack automation at scale?
3. How easy is it to resolve things like performance issues and bottlenecks?
4. Is it interoperable with other vendors in an OpenStack deployment?
5. Can it be deployed quickly and easily with a native OpenStack Cinder driver?
A lot of people forget that getting the storage layer right is essential for supporting your OpenStack deployment. While it may be the bottom of the stack, storage shouldn’t be your last consideration. And with a whole host of storage types and vendors to choose from, hopefully this list will help you narrow down the search and select the best storage layer to support your businesses’ automation strategy.