Enterprise Storage Forum content and product recommendations are editorially independent. We may make money when you click on links to our partners. Learn More.
Trade press articles might have you believing that virtualization tops the list of SAN killer apps. Paul Massiglia, co-author of Storage Area Networks – A Complete Guide to Understanding and Implementing SANS (Veritas Series by John Wiley & Sons) says, “Backup is the number one SAN killer in my book.”
Massiglia knows what he’s talking about. He has worked for major storage vendors for about 20 years. One of his key tasks as engineering technical director at Veritas Software includes representing the company at storage industry trade associations. His other task includes editor of the Veritas Series of books to be published John Wiley & Sons.
You just might call Massiglia a walking encyclopedia of storage. The Veritas Series will include an updated version of Massiglia’s earlier book, Highly Available Storage for Windows Servers. He also wrote Disk Storage Management for Windows Sever, which includes a primer on Veritas Volume Manager for Windows. During the time Massiglia served on the RAID Advisor Board, he authored The RAIDBook, a definitive technical handbook to this storage technology. He also acted as editor director of the Storage Network Industry Association’s Dictionary of Storage Network Terminology www.snia.org/resources/dictionary.html.
With seven more books to crank out, including one on disaster recovery, Massiglia took time to answer questions about his top three SAN killer apps, disaster recovery, and virtualization. Here’s what he had to say.
Why is backup the number one killer app?
Server backup has become the number one killer application for everyone. Think of backup as insurance for data protection. It doesn’t add anything to the bottom line, but don’t try to exist without it. Server backup comes with a host of problems, such as consuming too many application resources, and not being very reliable.
SANs provide for better backups by enabling you to connect all of your storage to all of your servers. All of your storage includes, not only online storage, but also your tape drives and tape libraries. Therefore, you can justify buying one expensive library and amortizing the cost of several servers. This technique gets the human handling out of the equation and provides for a more reliable backup.
A tape library will spit out cartridges that you regularly transported to an off-site underground vault. You can’t do that very easily with an EMC Symmetrix. On the other hand, Michael Ruettgers, EMC’s CEO, says that it takes six months to restore a petabyte. So, as the quantities of data the people store online get larger and larger, the entire issue of data protection gets more and more complex. When you get into the terabyte range, disk-to-disk backup – that enterprise disk subsystem to enterprise disk subsystem- becomes a necessary first-line of defense for data protection.
What are some of the other killer apps?
The second killer app is highly available data, which gets into the question about online backup. It also helps into the question of RAID and host-based RAID and RAID-protected data that can move back and forth between servers.
The third killer apps is what I call disaster recoverability. In other words, do you have RAID across the globe as far as data is concerned and as far as SANs are concerned?
What should you be doing for disaster recovery?
You first need to define what types of disasters you’re concerned about and what it’s going to take to recovery from them. For example, are you worried about an earthquake that’s going to take out a metropolitan area? Are you worried about a Mississippi River flood that could cut a swat through the middle of the country? Based on these events, you have to decide how far away from your main sites your disaster recovery sites should be. Next, you need to adopt techniques for getting data from your main sites to your recovery sites with the right balance of timeliness and cost.
Is virtualization one of your killer apps?
I put virtualization under highly available data and disaster recoverability. I’m spending a good part of my life running around talking about all the ways to do virtualization. Veritas has most of these bases covered. For example, we do host-based virtualization through Volume Manager. We also market a line of products that are called storage appliances. We sell it as software that you can put in a storage device you buy. It’s functionality approximates what DataCore offers by making intelligent boxes as part of the SAN, rather than embedding intelligent lumps of software both servers or storage subsystems.
When the dust settles, what will virtualization look like?
The jury is still out on that. In fact, you can’t be in the storage business unless you have the V word on your home page. Probably when the dust settles, virtualization will wind up something like caching. If you go back to the early 1990’s, you had many choices for caching. Do I cache in my servers, my adapters, or my storage subsystems? The answer is everywhere.
There are different advantages to where the virtualization takes place, however. People have been virtualizing in the RAID subsystem for about 10 years. A RAID subsystem firmware and hardware is a lot more robust than a Veritas Volume Manager is. So RAID can do a better job of virtualizing when the issues relate to disk drives. On the other hand, Veritas Volume Manager is adept at virtualizing across RAID boxes, especially across disparate ones. One of the reasons I might want to virtualize may come down to data availability or disaster recoverability.
Any comments on outsourcing of storage?
The storage service providers are in a mad scramble to get an industry started.
