Data Center Tech Blog
John Scaramuzzo

Senior Vice President, SanDisk Enterprise Storage Solutions

Let’s face it, we live in a data driven world. If you have seen any of the recent stats, you are likely as flabbergasted as I was when I read that we send 100 billion e-mails each day (just business email alone), upload 100 hours of YouTube videos every minute and text or share 500 million photos every day. Simply amazing.

It seems as though everything is moving to the cloud these days. In fact cloud data centers are growing fast with workloads in cloud data centers expected to grow five and half times more than traditional workloads by 2016 according to Cisco’s Global Cloud Index. As apps move to the cloud and people experience having access to their information at anytime and anywhere, the expectations are changing. Now, we all expect when we share a photo, watch a YouTube video, or connect to Salesforce, that the content will be available immediately. This expectation is placing a significant strain on IT data centers as they work to process and meet requests in near real time. This is causing IT managers to seek ways to improve total throughput and reduce latency, which is putting the crosshairs directly on storage.

So what do data center managers need to consider when addressing the storage needs of their cloud infrastructure? Performance is a no-brainer. You have to be able to push I/O at a high rate – otherwise you are facing the same issue you started with. However, it isn’t all about IOPS. Latency is also of supreme importance, and not just the need for it to be low, but consistent. Lastly, you have to make sure that your performance and latency metrics can endure, meaning you have to take degradation into account. SSD performance should stay as close to out-of-the-box levels throughout the drive’s life, otherwise you run the risk of your applications quickly screeching to a halt. A good metric to keep in mind for this is five percent, meaning anything that degrades more than this overtime should be thrown by the wayside as it will negatively impact the data center.

Beyond the performance, latency and degradation issues, when evaluating SSDs for your data center, it will be critical to plan for the worst case scenario when it comes to latency. Today’s SAS and SATA solutions offer respectable latency levels and PCIe-based drives offer an order of magnitude lower. While there are a lot of good options out there, many struggle to deliver fast access to information because they don’t have consistent latency. Their latency looks like a roller coaster and sees many spikes along the way. These spikes can be detrimental to your data center because the longer data sits waiting to process, the more time the user sits. And, when that happens, you lose their attention and likely the profits they bring. Our solutions deliver consistent latency – ensuring that your information is not sitting there waiting.

So, remember, when it comes to building an infrastructure for the cloud, think consistent, low latency, low-degradation and high-performance.

Join SanDisk® at Oracle OpenWorld, September 22 – 26, 2013 in San Francisco, CA

Brian Cox on theCUBE, Oracle OpenWorld 2013

subscribe blog