Spinning disk may be an attractive backup option, but the tendency to back up too much data and retain backup data for too long can limit hard disk drive storage's advantages.
"We're using disk unwisely. We're using half of the capacity that we deploy to back up. Why are we choosing to use disk for everything? Not every application requires disk," said Jon Toigo, founder of Toigo Partners International.
In this Storage Decisions video, Toigo discusses how to maximize the disk storage you have and make your backups more efficient.
Toigo said that organizations have to think about cost containment. He said that a typical solution for mirroring between two disk arrays is to use "copy after write," a function that writes to the primary array, and then another copy is created at a secondary copy of the disk array.
"Now, usually both those disk arrays have to be from the same vendor. And that secondary write is usually controlled by the array controller on array No. 1. It's a lock-in. For that configuration to work, you have to buy both kits from one vendor. That's a lock-in. That's going to cost you serious money," Toigo said.
He said the other approach is "copy on write," which means that when data is created, another copy is written to the target. "That can be done in software, and it doesn't really care what [brand] that second disk array is. … That is a better approach -- much less costly and just as efficient. In fact, in some ways, [it's] more efficient," said Toigo.
Toigo said he believed that virtualization is "the most cost-effective strategy I've ever found for disk-based replication." He said that a software abstraction layer placed over the disks in a storage array can be used to pull virtual volumes out of that aggregated disk, regardless of which vendors made those disks.
"Then do your mirroring from volume to volume. It doesn't care what the brand name is on the box. What you've done is add a virtual controller over the top of your infrastructure. I like this strategy," Toigo said.