Dedupe and Oracle RMAN Backups – it’s all about ROI
My last post walked through the details behind a 6 week Permabit Labs study of dedupe effectiveness with Oracle RMAN backups. To summarize, within a single week, using a standard regimen of full and incremental backups, we saw 10x data reduction rates – and that was with completely random generated data. When we extended out to 7 weeks of full backups we saw 16:1! What’s more, we were able to chart the increase in efficiency over time and demonstrate that it scales to much larger volumes. Put another way, our research proved that over a 7 week RMAN backup cycle, dedupe will effectively reduce 4 PBs of data to 250 TBs!
Even if I could manage to purchase an integrated 4 PB storage system today, at an industry average cost of $5/GB for enterprise storage I’d have to spend $20M USD just to acquire the hardware and I’d easily spend 4 times that once I’ve factored in the Total Cost of Ownership (TCO) aka power, cooling, floor space, maintenance and cost of administration for a total cost of $80M USD!
With dedupe, I can deliver the same data storage capability with 250 TBs of storage consumed – projected hardware cost, $1.25M USD and with TCO $5M USD. It’s certainly easy to see why dedupe is so popular for backup applications, that adds up to savings of $15M and that’s with uncompressible data!
These are great ROI numbers for any dataset. With database backups there has been a misunderstanding that they required specialized content scanning to deduplicate well. While Permabit has content scanning capabilities, our testing demonstrates that the general savings from popular block-based deduplication solutions (such as those typically found on primary storage systems) is significant for database backup environments.
Speaking of compression, did I mention Permabit just announced our new Albireo COMPRESS product? Next post we’ll talk about some of the results of our studies on the effectiveness of compression on databases.