Download the authoritative guide: Enterprise Data Storage 2018: Optimizing Your Storage Infrastructure
Next February, NASA Ames Research Center plans to launch the Kepler Space Telescope into an orbit around the Sun. Its 3 1/2-year mission: to look for not-so-strange, Earth-like worlds, to seek out new life, to boldly go where no telescope has gone before.
As for the scientists involved, their mission is to sort through the dozens (or possibly hundreds) of terabytes of data (images) the telescope will be capturing as it scans the Milky Way, looking at approximately 100,000 solar systems in 30-minute intervals, and beaming the images to the NASA Ames Research Center in Mountain View, Calif.
To ensure the integrity and preservation of all that critical data, and at the same time make sure it will be available to the many scientists working on the project, the Kepler team went in search of a reliable, scalable storage solution. Additionally, the project had strict budget constraints, so the solution NASA Ames chose had to be self-managing.
A Predictable Solution for Unpredictable Datahttps://o1.qnsr.com/log/p.gif?;n=203;c=204650394;s=9477;x=7936;f=201801171506010;u=j;z=TIMESTAMP;a=20392931;e=i
The first item on the Kepler team's storage must-have list was the solution's ability to store a large (and unpredictable) amount of data.
"Capacity was our number one requirement," said Chris Middour, deputy manager for the Kepler Operations Science Center. "Second was performance. It needed to be fast, because we're searching for planets inside all this data, and with 100 terabytes of data to rattle through, we needed it be able to retrieve very quickly. And number three was the cost of ownership."
Initially, Middour and his team looked at eight different storage solutions. Then they quickly narrowed down the choices to two, then one 3PAR's (NYSE: PAR) utility storage, specifically the 3PAR InServ S400 Storage Server, with up to 300 TB capacity. NASA Ames also purchased the 3PAR Virtual Copy tool for the project, which would allow NASA researchers to share snapshots of the images the telescope had gathered safely and affordably.
Middour chose the 3PAR InServ, he said, because of its "agility, resilience, and scalability" and because it would allow the project to stay within budget (at least on the storage side), thanks to its cost-efficient Serial ATA (SATA) drives, 3PAR Fast RAID 5, and self-managing capabilities.
"The cost of storage does not end at just the hardware, but includes also the staff to manage and maintain the systems," said Middour. With other storage solutions, he said, "we were looking at the need for a dedicated, full-time storage expert to administer our storage infrastructure. With 3PAR, we are counting on the system's ability to practically administer itself to spare us this additional expense."
So far, in just over two months of testing using software simulations, the 3PAR utility storage system is living up to expectations. "We're very pleased with it," said Middour. "It's very easy to use, and it works perfectly. It takes only a few seconds to take a snapshot of any size volume we can think of."
And, as advertised, Middour hasn't had to hire a separate storage administrator to manage the system. What little care or checking the system requires is done by a network administrator.
Of course, once the Kepler Space Telescope goes into orbit next year and starts beaming back terabytes of data, that could change, but Middour doesn't think so.
Meeting More Down-to-Earth Storage Needs
Craig Nunes, 3PAR's vice president of marketing, says you don't need to have 100 TB of data like the Kepler project to benefit from 3PAR's utility storage solutions.
"You can start with just a few terabytes and grow to an eight-controller system with loads of performance and terabytes behind it," he said. "Both [the 3PAR InServ S400 and the S800] can start with just a couple of controllers, a few terabytes and a couple of ports to connect your host. The 400 can then grow to a four-controller node cluster within our array [storing up to 300 TB], and the 800 to an eight-controller node cluster," storing up to twice that amount.
"It's that cluster that provides you with the back-end disk connectivity and host ports and iSCSI and performance all the things you need to get the job done," including self-management, said Nunes, which makes 3PAR utility storage (at least according to Nunes) "an exception in the block storage world."