HPE Storage Users Group

I think the concern is that a leak doesn't constitute an announcement. And particularly with something like flash cache, one would expect more technical details of the implementation.

Sans an official announcement. Hp could just change their minds and not release it

This HP document used to contain the leaked details about Adaptive Flash Cache coming in 3.1.3 MU1, but has since been removed. Hopefully, thats not a sign of delays and just policing up the rumor mill.

http://h18000.www1.hp.com/products/quic ... 64_div.pdf

_________________
Richard Siemers
The views and opinions expressed are my own and do not necessarily reflect those of my employer.

Thats what im worried about.

There's been a lot of silence surrounding a feature that would be a great competitive point when they are competing against emc/netapp.

That leads me to believe it'll be awhile before we see it yet

Adaptive flash cache (the way I imagine it would operate) would suit our environment perfectly. We have steady loads but at times we must deploy new VMs and copies of existing LUNs that exceed the 64 GB of RAM cache that is in our 7400 arrays. When we deploy new eager zero thick VMDKs to TPVVs we experience extreme latency, but only if we deploy new VMDKs exceeding 64 gigabytes.

Our local HP reps have proposed more SAS drives but has also alluded to adaptive flash cache, but with new technologies being announced, there is only one concrete solution to our issue, which is buying more SAS drives. After I heard about adaptive flash cache, I asked if that it would be wise to invest in an SSD tier in the immediate future and forgoe the large chunk of SAS drives they suggested a few months back.

I'm going to leave this on a cliffhanger and you can look forward to hearing from me when I have more to tell.

From TheRegister.co.uk - http://www.theregister.co.uk/2014/06/18/las_vegas_3par/

You know what they say: What happens in Vegas, goes on the internet – right, HP 3PAR guys?
Rumours trickling into our aural organs
By Chris Mellor, 18 Jun 2014

Blocks and Files Hush there, 3PAR fans, and know of the rumours whispered into my ear following HP's Discover confab in Las Vegas this month.

Here we go:

We’ll see a better unified management console and new Remote Copy. We heard about triple data-centre replication with zero data-loss; async streaming for Remote Copy to give EMC's SRDF a tremor in the testicles; and async periodic Remote Copy.

Our super sensitive ears heard faint conversations about Adaptive Flash Caching (AFC) which uses the flash released by Adaptive Sparing as a cache, perhaps caching frequently accessed data from the controller cache. That should avoid having to buy dedicated SSDs to do that caching work. AFC could be included with the 3PAR OS and might even be called Accelerated Performance, taking 3PAR to the (v)max.

3PAR could get object storage and file storage added to its block storage to become a converged and unified data store by the end of the year. How FAStastic will that be? Won’t NetApp be upset, and won't VNX fans be fulminating?

We also heard faint murmurings of an Eagle, an Apollo chassis with Ivy Bridge servers and disk shelves filling its ten slots. Cascading disk shelves were mentioned, whatever that means.

You know what they say; what happens in Vegas stays in Vegas. Oh, no it doesn’t.

If true, hopefully lots of interesting things coming - but probably still under NDA largely.

When Carnarvon asked "Can you see anything ?", Carter replied with the famous words: "Yes, wonderful things."

Some more stuff on the around-the-storage-blog about "how it works..."

http://h30507.www3.hp.com/t5/Around-the ... a-p/164858

I was thinking too - this $2/GB claim/marketingpseak - might only be when the de-dup is used - if the 4:1 average saving is achieved - then they are really claiming you pay $8/GB but get 4x the data into it, thus averaging $2/GB.

Yes it includes the compaction savings just like all the other flash vendors claims.
Dedupe testing has delivered between 4:1 and 10:1 so the 4:1 appears conservative at this point. The numbers also include the additional space unlocked from adaptive sparing. On top of that hopefully you''ll make additional space savings with 3PAR's established thin technologies.

So I've heard some interesting things this week.

Firstly, the 7200 and 7400 are also going to get dedupe, not sure whether this will be limited to ssd cpgs.

Secondly that flash cache is coming in September, it will be a minimum of 4 ssds to configure it, if you already have ssds you can use a portion of the space on them and don't need to add more ssds.

Thirdly, there is a new simpler to configure quorum coming for peer persistence, along with support for hyperv on pp too.

Whether any of this comes to fruition in September remains to be seen, but it's pretty much what we thought would happen anyway.

I will believe flash cache when I see it. It was due last December and missed. Also from what I have been told flash cache uses only a small amount of space and that varies whether you have a 7000 or V class. It is not like if you add 10TB of SSD that it will use it for flash cache.