You’ll want to ensure you choose a storage platform that can scale to meet your capacity demands over the lifecycle of the infrastructure. Another approach here is storage pods, or infrastructure pods that include the storage. With a pod approach you purchase the storage or infrastructure in chunks based on performance service-level agreements, such as IOPS, number of VMs supported, and so on. When those capacities are reached, another pod is purchased. They key to this approach is an automation and orchestration layer that can manage multiple pods.
Performance is always vital to success of any storage deployment. In a typical private cloud deployment, you’ll have multiple tiers of performance levels, commonly defined in terms such as silver, gold and platinum. You’ll want to ensure your storage platform can handle multiple service levels without breaking the bank (that is, you don’t want to be paying for the top tier for every service). Storage features such as auto-tiering (the ability for the array to move frequently used data to faster disk, and vice versa), and cache (front-end cache memory banks) allow for higher performance with lowered costs. Additionally, features like deduplication, replication and snapshot technologies should all be weighed and considered.
The last key factor to discuss is manageability, which plays a large factor in the TCO of a storage deployment. For private cloud deployments, you’ll want a system that can be easily managed and monitored both with native tools and through the automation/orchestration platform you choose. The more open the storage management platform is, or the more robust its API or software partner ecosystem integration is, the better off you’ll be.
When making this decision, make sure that the software products you’re considering for the automation/orchestration platform can manage your chosen storage platform well. These tools will be the most commonly used and can be quite expensive, so it’s important not to limit your options with hardware decisions.
Overall, the storage decision is a complex and important one when it pertains to private cloud. Even if you’re just looking to virtualize at this time, you’ll want to consider the implications of the storage choice on potential future cloud initiatives and weigh your options carefully. Where possible, rely on proof of concepts and customer references in similar deployment scenarios--because everything looks great in PowerPoint.
Disclaimer: In my primary role I work with several storage products and vendors; this article is not an endorsement of those products or vendors.Joe Onisick is the founder of Define the Cloud and a principal engineer for Cisco's INSBU. Onisick has 17 years of IT experience spanning a broad range of disciplines, starting with server and network administration. From 2000-2005, Onisick was a US Marine, where he served in ... View Full Bio