Hu's Blog

Awhile back I posted a blog, Data Replication: When push comes to pull, where I described our Hitachi Universal Replicator, as a pull technology for asynchronous replication. With HUR the remote site USP pulls data from the production site USP based on a time stamped journal. The benefit being the reduction of resource consumption at the production USP, by transferring the majority of the workload for asynchronous replication to the remote USP. I compared that to feeding my daughter when she was a baby, where I had to "push" the food versus feeding her when she was older and able to feed (pull) herself.

I also extended the analogy to the present time where she is an adult, living away from home, and able to cook for herself. At this time I only need to send a recipe to her to recreate the same meal that I cook at home. In other words, I was alluding to a future time when we could do replication of data without having to send all of the data to the remote site. I would only need to send meta data to recreate the data. Wouldn’t it be great if we could replicate 250TB of data at a remote site, thousands of miles away, without having to actually send the entire 250TB! 

This week HDS announced the availability of ProtecTIER deduplication which can be combined with our Universal Replicator technology to make this a reality.

ProtecTIER has a very efficient indexing schema which enables a petabyte of data to be indexed by 4GB of memory. The power of this index has been shown to provide around a 25:1 reduction in stored data due to deduplication of redundant data patterns. This is not only a tremendous savings in storage cost, it also reduces the cost and effort of replication.

Now I can replicate 250TB of data to a remote site, by sending only 10TB of deduplicated data, a tremendous saving in bandwidth and an even more tremendous savings in our most valuable resource, time!

Sending deduplicated data with HUR is like sending a recipe to my daughter.