Version: 2008

Comments on: Deduping: Killer app behind battle for Data Domain

Process known as data deduplication is white-hot. So hot that Data Domain, virtually unknown outside the storage world, is now deemed worth at least $1.9 billion.

by bj1126 June 12, 2009 1:14 PM PDT
I am watching this one closely. It will be interesting to see what comes of this.
Reply to this comment
by JohnSWebster June 19, 2009 9:08 AM PDT
It was once the case that dedupe was only for backup. Then NetApp allowed users to dedupe primary data stores under ONTAP. Other vendors are following suit. There will be a number of variations on this theme for different applications I believe. I first encountered dedpue when I met a company named Rocksoft back in 2003. Their patents for a process they called Data Coalescence were filed in 1992. They were bought by ADIC, which in turn was bought by Quantum. If you're watching this space closely, that should ring some bells. JW
by Seaspray0 June 14, 2009 7:21 AM PDT
You need to look at the way they shrink data. Basically, they do what's known as a "single instance store" of data. If you have 10 GB of data, the first full backup takes up to 10 GB (compression usually gets you about 30% savings). The next time you do a full backup, the app sees that you're backing up the same data (sectors are identical) and only puts in pointers to the first data. So... your second full back of the same data takes a very small fraction of space since you only retain one single instance of data and point to it over and over again. If your data is fairly static, you can do hundreds of backups and consume roughly 11 to 12 GB of space (one instance of data and lots of pointers). On tape, you would be consuming roughly 1000 GB of space. "Cool!" you would say, and yes it is. But there is a danger as well. Each tape holds its own instance of the data. For 100 tapes, that would be 100 instances. Redundant? Yes it is. But I'm also concerned of only having 1 instance being referenced 100 times. All is takes is one failure in one instance, and the data is gone for all 100 full backups. For that reason, Data Domain doesn't use raid5. They use the next revision of the raid where you have 2 parity drives per logical disk instead of one (allowing 2 disks to fail and still retain the data). I would prefer to include a mirror of that as well. I want more than one instance. Data Domain gives you the option of duplicating to another appliance and this satisfies my desire, but the cost is double for that solution. Adding the second appliance makes tape very competitive in price. I will still keep an eye on the cost. If Data Domain lowers their prices where doubling up the appliances is cheaper than tape, I'd very much consider it.
Reply to this comment
by JohnSWebster June 19, 2009 9:17 AM PDT
While I?m not going to argue the case for DataDomain or their pricing practices, I will say that with higher density disk on the near-term horizon, the cost per GB will likely come down to the point where your analysis tells you it?s time to jump in. Personally, I think we?re at that point now, which is why there is the huge interest in DataDomain. JW
advertisement
Click Here

15 sites that went kaput in 2009

Web sites launch all the time, but they also shut their doors. We highlight 15 that bit the dust this year.

Top 10 news stories of the decade

Let the debate begin: Was the iPhone more important than iTunes? Was anything bigger than Google finding a great business model? CNET offers its list of the 10 most important stories of the '00s.

About Data-driven

Storage is more--way more--than a mere peripheral. In Data-driven, John Webster probes into storage technologies, the vendors behind them, and how customers use them in the context of market drivers such as Web 2.0, cloud computing, and the need to get meaningful information from the data fire hose that is now part of our daily life.

John is a senior partner at Evaluator Group. He has served as principal IT adviser at Illuminata and has held analyst positions at IDC and Yankee Group Research. He also co-authored the book "Inescapable Data Harnessing the Power of Convergence." John is a member of the CNET Blog Network and is not an employee of CNET.

Add this feed to your online news reader

Data-driven topics

advertisement