When talking to people who don't have a strong technical
background, I find that their reactions prove that Arthur C. Clarke was right
when he said, "Any sufficiently advanced technology is indistinguishable from
magic." To them, any technology that lets them store 10TBytes of data on a 1TByte drive
is magic. Then I start to explain to
them that I can't really predict how much their data will dedupe because it
depends on the data and that "your mileage may vary," and they start wondering if it's
magic or snake oil. With BridgeStor's
new deduplication simulator, we can finally predict how much reduction you can
get with your data.
BridgeSTOR's VS-ADR (virtual storage-advanced data
reduction) simulator will scan your real production data and determine how much
it would shrink if you stored it on one of BridgeSTOR's appliances. At the end, it creates a report showing how
much data reduction you can expect to see on your data and how much of the
reduction is due to deduplication and compression. It even comes with an Excel spreadsheet that
generates pretty graphs that you can show your management to convince them that this
magic stuff is worth having.
Of course BrideSTOR is giving the simulator away to help
it sell its deduplicating storage appliances. BridgeSTOR founder John Matze also founded iSCSI
vendor Siafu, and he was a key player in the development and marketing of HiFn's
hardware assist technology for deduplication, compression and encryption. At
BridgeSTOR, he's combining those technologies into HP server-based appliances that are preconfigured
and optimized for Windows Storage Server, Data Protection Manager and Backup
Exec.
The VS-ADR simulator installs on a Windows system and will
analyze the data in a \SCRATCH folder on whatever drive you point it at. BridgeSTOR says it will process about 28GBytes of
data an hour on a modern server and about half that on a laptop. As always, your mileage will vary based on the
processor and storage system you choose. Since it's running in software, and since it's free, we have to assume
that it's not terribly optimized software, and since the real BridgeSTOR appliances use
hardware assist, the simulator's performance shouldn't be used as an indication
of how fast a real appliance will reduce data.
While I'd have preferred to be able to point the simulator
at my production data, I'd probably be upset when it took five days to process
several terabytes of data and loaded my production systems with all that
I/O. All in all, I'm grateful to have a
tool that will help me estimate the data reduction I can get on real-world data
and will have to stop looking a gift horse in the mouth.
