The perfect server flash storm hitting storage arrays has generated EMC's well-signalled Lightning strike; VFCache has arrived, extending FAST technology from the array to the server. Project Thunder is following close behind, promising an EMC server-networked flash array. This is a major announcement and we are covering it in …
How would this work in a vSphere cluster or any cluster for that matter? E.g. 3 hosts online and 1 host loses its connection to the remote block level storage device - what happens to the "in-transit" data which is sitting on the flash card. Is there data loss experienced or is there software in place that allows the cards in each host to communicate with one another and somehow offload the data? Also, is there any 'real' value in this given the current amounts of memory available on servers these days + 16GB FC + SAN cache + SSD drives??? I am yet to see an app that "requires" sub 1 milli-second response times - read or write. (I am seeing <=1ms already with these other caching-technologies)
This is an interesting development.
About half a decade ago, I designed a framework for an ecommerce site based up a similar architecture. Each web box had a sql server express sitting on it, and when data was pulled across, it was cached there, and then a job, based on the timestamp, checked for updates to the local data. The ultimate research showed that each web box, ultimately could simulate a personal shopping assistant, and it turned out to be monster scalable, but you had to design in the kinds of inefficiencies that shopping assistants have evolved, such as shouting out, "Jenny! have we got any more penis enlargers coming in, this teenage boy says there's none on the shelves."
It worked quite well in the main, but added a layer of complexity that the client at the time couldn't afford (an extra developer for a few months was just too expensive.)
It's good to see the opposite of lean supply chain working here, as it doesn't cost much to over order stock when the stock is just data.
I suspect we'll see more of this approach in future, although, it's a fix using software design, rather than a physical technical improvement. No reason why we can't have both.
A few remarks on parts of the article:
EMC's FAST (Fully-Automated Storage Tiering] is not new in the industry, Compellent (now owned by Dell) did it already with Automated Tiered Storage since 2004
Also what makes me sad, is there is still the focus on FC while there are other technologies available (10 GbE, FCoE, etc.) that can do the same but for a lower price.
Its a interesting article, lets see what the competitors come up with :)
New Kids on the Block ....... and Not in Any Way the Same as that which is replaced.
"The threat here is that primary data could move from networked arrays into direct-attached server flash storage. ..... Chris Mellor
You may like to consider that others can deliver that as a promise, and the threat is to those who would be exposed as being formerly instrumental in benefitting exclusively from the sensitive incriminating data which such control of networked arrays delivered.
Hmmm, so the data could move into direct-attached disk storage? So kind of like an internal drive then.
There's a reason why data is on centralised systems. It allows for more efficient use of drives, and gives you extra features, in particular replication, consistency groups etc., which require central storage systems.
Of course, there's no reason why reads can't be serviced from something closer to the CPU (like server cache already does). This is all that's happening here, and it's a good idea if you want low latency reads.
does anyone edit these articles?
I'm not by any means a perfect typist, but this article is just plain painful to read. Flick the grammar checker to "on" please.
After watching the opening few minutes of EMC's VFCache superhero BS (http://developer-content.emc.com/community/connect/live_event/lightning_feed_640.htm), then sifting through your typos, I've had enough. Beer time, and it's only 2pm!
- Mounties always get their man: Heartbleed 'hacker', 19, CUFFED
- Analysis Oh no, Joe: WinPhone users already griping over 8.1 mega-update
- Leaked pics show EMBIGGENED iPhone 6 screen
- Opportunity selfie: Martian winds have given the spunky ol' rover a spring cleaning
- OK, we get the message, Microsoft: Windows Defender splats 1000s of WinXP, Server 2k3 PCs