Published: 12 Jan 2008
| Backup and restore environments keep growing, and every one terabyte of production data typically translates into many more terabytes of backup and archive data. While backup environments consume scads of disk capacity, most traditional disk storage systems offer limited scalability. Buying another storage system can be the only recourse for companies that max out disk space.
So what do you do when you have a petabyte (PB) of data to back up? If your backup workflow involves a huge volume of data and many servers, doing backup during off hours can turn into rolling backups over multiple days.
But if you can infuse your backup system with enough performance, "you can reduce or eliminate the need for rolling backups, and back up your entire environment in one night," says Josh Goldstein, VP of product marketing at DataDirect Networks. The company's Silicon Storage Architecture (S2A) Appliance, a primary storage system, is also used by companies such as AOL for high-speed data backup and restore.
Companies are also considering alternative architectures for better secondary storage performance, such as the Hydrastor grid storage system from NEC, which wasn't specifically designed to handle primary storage apps. Generally available since last October, Hydrastor offers unrestricted scalability through Accelerator Nodes, which scale performance from 200MB/sec to 14,000MB/sec; and Storage Nodes, which scale disk capacity from 150TB to 10PB as one logical pool of capacity. (Nodes are NEC's term for industry-standard servers.)
"If you need to scale, you just stick another node on a rack," says Karen Dutch, NEC's general manager, advanced storage products group. "There are no LUNs, no RAID groups, nothing that you have in a traditional system."
Heidi Biggar, an analyst at Enterprise Strategy Group, believes the Hydrastor unit has a lot of appeal for companies with high-capacity backup environments. With many disk storage systems, she says, "the only way to scale is to buy another system." Companies using multiple disk-based systems could "stick in one Hydrastor and then add to it," she says. "It's easier to manage one system than multiple systems."
DataDirect's S2A Appliance is designed to be several times faster than generic SAN RAID systems. The company says the system delivers sustained real-time performance of up to 3GB/sec on writes as well as reads.
"The key is the array's write performance," says Goldstein. "A typical RAID storage system cannot write to the disks as fast as they read from them, so if you have a write-heavy workload, like backup, you need more of those RAID systems to gain that throughput."