How object storage data protection differs from file system backupDate: Oct 22, 2013
"It's a file, it's a piece of a file, it might be an email, might be a picture, it could be a number of things," said Preston of an object. "It's basically an entity that we can individually address or refer to. It really depends on which vendor you're talking about."
Preston said the storage process for objects is very different than the way we write data in a regular file system. Object storage also can be "coarse grain" or "fine grain" -- the latter of which he described as taking a file and slicing it up into chucks. The resulting chunks are objects, he said.
"Each thing that I'm going to put into the system -- whether it's an email, or a file, or a piece of a file … is given a unique address and that address is simply a string of digits that is comprised of the content of that file. And any other object would have a different address because it has different content," Preston said.
Each object is assigned a hash, which is used for location and for deduplication, and which allows the system to discard duplicate objects, he said. He went on to explain that in object-based systems, when a piece of media fails, it can be rebuilt because the system has replicated objects elsewhere on the array.
"You can even do object-based replication across the WAN," he said. "So you can say, 'I want each object to exist in at least two locations [locally] and two geographic locations, as well.'"