" Because flash arrays are so much faster, the college has found a 57 per cent reduction in its storage requirements" - huh? Since when has speed anything to do with storage space?
Manchester college swaps out disk for rackful of hybrid flash
A hybrid flash/disk array is replacing several disk arrays at a UK college and shows the way mainstream disk array replacement is heading towards flash for primary data applications. Bolton College in the UK has replaced eight Dell EqualLogic disk arrays used for databases, virtualised servers and virtual desktop environments …
COMMENTS
-
-
-
Thursday 16th March 2017 13:00 GMT Stuart Castle
Maybe.
It's also possible that various departments had stuff cached locally, to increase performance.
Of course, it's also possible they had never properly planned their storage systems, merely letting the systems grow as needed, and as a result, never implemented a proper deduplication system, so had duplicate copies of a lot of data for that reason.
-
Thursday 16th March 2017 15:06 GMT John Brown (no body)
"never implemented a proper deduplication system, so had duplicate copies of a lot of data for that reason."
That's what I took from the article. Very large amounts of data take time to dedupe, especially across 8 separate arrays of spinning rust. Deduping on flash all in one box and using the flash to assist the dedupe of the spinning rust makes it much fast and more manageable without hitting on the array performance.
-
-
Thursday 16th March 2017 15:56 GMT twister68
Performance = Inline Data Reduction Services
Actually for the record (excuse the pun) the superior performance enabled inline compression and inline deduplication the result of which is far reduced physical capacity and a performance multiplier.
No over provisioning required. Perfect use case for AFA & Hybrid in same Array.
Twister
-
Thursday 16th March 2017 21:39 GMT Anonymous Coward
Re: Performance = Inline Data Reduction Services
There's nothing magical about flash that makes it better at dedupe and compression. It is just that because flash cost a lot more when all flash arrays were first being designed, they were forced to invest more money in the controller to make it capable of dedupe/compression to keep the data requirements down. If they used the same controller in a disk based array it could get the exact same savings.
-
Saturday 18th March 2017 20:14 GMT Alan Brown
Re: Performance = Inline Data Reduction Services
The killer limitation on dedupe isn't flash or rust, it's ram and to some extent CPU horsepower. Deduplication is a 2^n problem and older controllers with tiny amounts of ram simply couldn't cope with large amounts of dedupe.
There's a lot of discussion of the problem in ZFS fora - the concensus is that dedupe is generally best left OFF except for specific uses in small areas such as mail spools or document stores.
-
-
-
-
-
Thursday 16th March 2017 13:19 GMT Anonymous Coward
I thought that equallogic arrays were 'scale out' in that you add a 'whole array' each time to the current array, each disk cabinet was an array (controllers etc) but they linked to the other arrays in the array cluster, and were managed as one unit from the same interface. I have never used one, but that was what was presented a few years back by dell when we looked into them.
-
Thursday 16th March 2017 18:55 GMT robertcirca
A single array ...
I am quite old and so is may brain.
However, I still remember that a university in London had data problems recently.
A single array ??? (a thousand more "?"s).
What about replication. Do they have a disaster recovery plan. And just in case they do, I bet they have not tested it.
The system was probably designed by an outsourced sales SPECIALIST from India.
Good luck, very good luck.
-
Thursday 16th March 2017 19:32 GMT Danny 14
Re: A single array ...
My thoughts exactly. We use about 20Tb and i went for 2x SANs that could replicate in different buildings as i was paranoid (backups are separated again, this is just for the cluster nodes and sans).
No way in hell id run a single point of failure and we only have about 2k users not 13k.
-
-
-
Thursday 16th March 2017 23:42 GMT Anonymous Coward
We had a (brief) look at Tegile to replace our aging Netapp (4 year old controller pair, and the oldest shelf dates back to 2009?), and while it was certainly impressive, the cost was... not. (the purchase cost was more than renewing the maintenance on said netapp by a good margin, and slightly less raw capacity.)
We ended up going to a nimble CS500 with a pair of the H90 expansion shelves; it was about 2/3rds the cost vs the renewal, a good deal more capacity, and easily the same performance figured as the netapp.
Anon to protect my paycheck.
-
Monday 10th April 2017 06:28 GMT Anonymous Coward
Erm?
Is this an advert?
I guess this is why Tegile is just about to be consumed by its major shareholder, the fact that the best case study it could find to shout about is a small college in a small uk town.
Must have been a slow news day for thereg too if this non news story made it to the page.