Essential guide to hybrid flash arrays
A comprehensive collection of articles, videos and more, hand-picked by our editors
Although some consultants and industry analysts advise enterprise IT shops against purchasing solid-state drives...
By submitting your email address, you agree to receive emails regarding relevant topic offers from TechTarget and its partners. You can withdraw your consent at any time. Contact TechTarget at 275 Grove Street, Newton, MA.
(SSDs) on the open market and self-installing them to save money, the do-it-yourself approach (DIY) to SSD installation is feasible in select situations to boost performance in server and storage environments.
The DIY approach becomes a consideration for IT shops due to the substantial price markups for SSDs by server and storage vendors. OEMs typically hike the price two to five times above the price of the original supplier's component, according to Jim Handy, chief analyst at Objective Analysis in Los Gatos, Calif. But even experts who say it is safe to buy and install SSDs yourself recommend more diligent assessment and testing before beginning the project than you would do with more familiar hard disk drives (HDDs).
You should also be aware of the differences among types of solid-state storage, such as single-level cell (SLC), multi-level cell (MLC) and enterprise multi-level cell (eMLC) flash. Wear-out projections and reliability differ according to type, with SLC drives lasting longer and running more reliably at a higher cost.
Dennis Martin, founder and president of Arvada, Colo.-based Demartek LLC, a computer industry analyst firm with its own on-site test lab, noted that his company bought MLC SSDs and installed them as boot drives in servers and desktop computers. Martin said the DIY practice is fine for SSD installation after an IT shop calculates its budget, determines its capacity and performance needs, and targets the appropriate workloads or scenarios. You should understand ahead of time that you will get less capacity but more I/O per second (IOPS) per dollar with SSDs.
"Any storage you can do with hard drives, you can do with SSDs," Martin said. "If you're comfortable with getting in there and putting in new hard drives, then you should be equally comfortable doing it with SSDs."
Sergis Mushell, a principal research analyst at Gartner Inc., said he has had many conversations with clients considering the DIY approach for SSD installation. But Mushell advised them to carefully assess the IT force's skill sets and time resources, the data sets that need a performance boost, and the backup and recovery plans for the data.
Mushell said that while "anything is doable given the right skill sets," he advises against a DIY strategy at most IT shops for SSDs. "As this technology is developing, there are many unknowns still that could cause risk within the enterprise, and therefore, it is best to go with known solution vendors or wait until the product has come down to a price that's tolerable for that enterprise for that application," he said.
Benjamin Woo, managing director at Neuralytix Inc., an IT strategy consultancy in New York, views the DIY practice as "irresponsible" for most enterprise IT shops. He said there are exceptions, such as Amazon, Facebook and Google, which are "in the business of computer engineering." He said he also knows of a major U.S. investment bank that engineered its own SSD-based infrastructure with thousands of servers using commodity parts.
But Woo said most enterprise IT departments will find the potential upfront cost savings negated by the increased need for maintenance and support. He wrote in an email, "Even leading storage/IT vendors spend hundreds of man hours to optimize and stabilize SSDs. Unlike contemporary HDDs, SSDs still need significant amounts of engineering to ensure that the [devices] are working optimally."
IT shops should always test the self-installed SSDs in a non-production environment before deploying them in production, said Marc Staimer, founder and president of Dragon Slayer Consulting in Beaverton, Ore.
"You need to protect yourself. You need to know what's going to work and not work, and be sure that everything's compatible," he said. "The worst place to find out that it doesn't do what you need it to do is after you've deployed it into production."