Analysts such as Greg Schulz of StorageIO are often asked by clients whether they follow their own advice, and as Schulz puts it, when it comes to backing up data, "I do eat out of my own dog food dish."
Schulz started on the customer side, working in the data center while working his way through school in operations. Then he moved into business development and back into operations designing backups, server storage, "and all those fun things." He has worked as a vendor selling software as well.
"I do talk about it, but I implement it for my own business and also for my personal matters. I walk the talk. I try things out. I learn from mistakes," Schulz says. "We need to learn from our mistakes. Don't be scared about making a mistake. The best thing to do is find out you made a mistake and fix it before it catches you."
Schulz explains that he recently had a file he needed to access, but found out it was corrupted. He went to his on-site backup copy, and found it was bad. "I went to my on-site backup backup copy. Well, it was bad. This gets funny. People tell me, 'Greg, you keep too many copies,' but I went to my cloud copy. Guess what? It was bad.
"What I was able to do was say, 'Time out, Greg. Think about it. Go back in time. Find out where the file was. Is it an isolated file [that was corrupted] or was it widespread?' I saved the wrong version. It was my fault. It happens all the time.
"You want to have multiple copies. You want to have multiple versions in different places. Long story short, this should be Data Protection 101-type stuff. By having those different copies in different places, different versions, I was able to go back far enough in time, get the right one, put it in place and everything was restored."
Schulz says that he protects certain data more frequently than others. "There's some data that is very static, where I may only protect it once a month, but I may also have multiple copies in different locations, but it doesn't change, so why am I protecting it? There's other data that I'm protecting throughout the data. My email is very critical, so I keep it small, but I'm protecting it constantly. Other key files, same thing: They're being synced and replicated, and I'm doing discrete versioning as well. Whether you are doing snapshots or versioning, you want those different time intervals and different versions you can go back to, both local and remote, so you can find a place in time and go back to it."
Schulz references the 3-2-1 adage when it comes to backing up data. "Keep three different copies of something, in at least two different locations, one of which should be off-site. I say no, it should be 4-3-2-1. You should keep at least four versions, three copies, two different locations, one of which at least is off-site.
"I protect to the cloud and I protect local, because there are times when I can't get to the cloud and there are times when it's quicker for me to go to the cloud than local, but it's a balancing act if I need to bring everything back quickly. It's faster for me to go get removable drives off-site, bring them back, stream them on, and go to the cloud and get the most recent updates."
Schulz has simple advice for admins worried about backing up data: "Change how you're protecting. Change when you're protecting. Don't treat everything the same. Instead of trying to brute-force it, stretch your dollar, stretch your budget and then use the different tools in new ways."