back to article HP hopes to give itself a shot in the arm with its LeftHand

We've met StoreOnce, HP's in-house deduplicating backup products, and the StoreVirtual virtual storage appliance; now HP is launching its range of StoreVirtual and StoreEasy hardware and software. StoreVirtual is a brand name for the P4000 LeftHand iSCSI storage arrays and not just the virtual storage appliance (VSA) …

COMMENTS

This topic is closed for new posts.
K
Bronze badge

Not enough...

HP are always behind the curve on this.

I'm in the market for some new arrays, budget is not such a concern, and I've evaluated the Dell EQL, HP Left Hand and EMC VNX. We've decided to run with the VNX simple because of the functionality it offers such as the SSD Caching and Tiering.. Dell offer tiering on their high end EQL, but AFAIK HP have no such functionality for arrays in this price bracket.

2
1

Re: Not enough...

Dell fluid data tiers across arrays too, so you can buy a SATA array and a SAS array, throw them in the same EqualLogic group, and it will redistribute hot blocks across the faster array.

HP's product suffers from two things, if you want a faster tier of disk you have to go out and buy new LeftHand arrays and manually move the data (AFAIK), and if you buy a starter kit with two arrays, you have to use one of them as a mirror. So a 14.4TB kit with two 7.2TB arrays gives you less than 7.2TB usable (with sparing). If you want network RAID-5 you have to add another array, and it has to be a specific part number to match the other nodes.

14.4TB in a network RAID-5 configuration (1 2-node starter kit and 1 additional array, all SAS) rings in at about $85k. A 14.4TB SAS EqualLogic array will cost me $35k, a 13TB SSD/SAS hybrid array rings in at $45k - I can buy both for less than the LeftHand stuff and get more storage capacity with automatic tiering to a higher performance array. Both quotes are on my desk as I type this (I think Dell wants it more than HP from the discount levels being offered).

HP actually offered up even stronger discounting on 3PAR, which is in a whole other class of performance altogether.

1
0
Anonymous Coward

Re: Not enough...

Network raid does have an overhead, but it also allows you to do some very clever things that just aren't possible with other architectures, it also provides more availability out of the box which continues to scale along with performance as you add more nodes, just the opposite of Equalogic. Fault tolerance and availability have always been about sacrificing capacity, raid overhead, spares etc, Yes you do lose usable capacity, but that capacity is relatively cheap being based on industry standard server hardware,

Equalogic is not known for it's efficiency once you start using snaps or replication, both of which have a huge block sizes (MB's) in comparison to Lefthand so any perceived usable capacity advantage can disappear rapidly once you start to use features. As for striping across Dell Equalogic boxes, it's not really recommended, lose one of the arrays in that configuration and you can kiss goodbye to your data. That's why you see a lot of shops with many Equalogic arrays all being managed in isolation, that's the reason Lefthand has Network Raid, something I'm told Equalogic are also working on, in which case they know they have an issue otherwise they wouldn't be looking to fix it.

The more margin you have in a product the deeper the discount you can sustain, that's why you'll see deeper discounting on higher end gear, maybe your reseller isn't passing the discount on ? but I do agree LeftHand needs sub LUN tierig even if it's just to put a tick in the box.

0
0
J.T

Re: Not enough...

They're getting desperate to quote 3PAR for 14TB.

0
1
Anonymous Coward

Re: Not enough...

Speak to your HP Sales rep straight away if you want to know what you should buy instead of the VNX!

http://h17007.www1.hp.com/us/en/storage/nextera/index.aspx

0
0
Anonymous Coward

Re: Not enough...

Well maybe that 'whole other class of performance' is about to become much more affordable!

0
0
Anonymous Coward

Re: Not enough...

Take it from this storage consutant:

VNX tiering has too many exceptions and block sizes are too large. If you want good tiering, take a closer look at EQL, Compellent or 3PAR. The SSD is just a red herring and you likely have no need for what this is offering anyway, but if you do, you can get SSD in the above 3 models too and your TCO will be lower.

2
0
Anonymous Coward

Re: Not enough...

That would be the long awaited P7000. One hopes they've worked out the ASIC reliability issues by now...

0
1
Anonymous Coward

Re: Not enough...

What ASIC reliability issues ?, surely EMC aren't still trying to pedal that same old FUD. I see more and more desperate attempts around this every week, to the point that it's become a bit of a joke even for Customers. 3PAR's obviously causing you some issues.

1
0
Anonymous Coward

Re: Not enough...

ASIC reliability issues, what you mean like this ?

http://www.theregister.co.uk/2012/01/13/tieto_emc_crash/

0
0
Anonymous Coward

Re: Not enough...

All the EMC FUD helped me out a treat the other week - I was up against them and they were feeding a load of the usual nonsense to the customer. Once I'd proven just how farcical some of their claims were, they got thrown out! Thanks EMC.

1
0
Anonymous Coward

Re: Not enough...

Netapp are doing the rounds with their non record breaking SPC-1 results as well. They show the Customer the latency at the end of the 3PAR test run (100% load point) versus their own 6 node version that artificially stops the test run before they become CPU bound and their own latency begins to rise. Of course there's no mention of them only doing a little over 1/2 the IOps.

0
0
Anonymous Coward

Re: Not enough...

You should take a look at V7000. The real time compression is the killer function on that array. Right in the same range as VNX, HUS... a tier above EQL, Lefthand and the iSCSI stuff.

1
0
FAIL

No thanks

This would be the same LeftHand OS that's had two grave remote-root buffer overflow vulnerabilities, which HP's own security researchers couldn't get the product team to patch in less than a YEAR:

http://www.zerodayinitiative.com/advisories/ZDI-12-166/

http://www.zerodayinitiative.com/advisories/ZDI-11-111/

HP's response to customers on both of these was: "Well, 'best practices' are to install the nodes on an isolated VLAN, so there's no problem and we'll be patching it when we get around to it."

We decommissioned our last LeftHand nodes and broke them down into their smallest components, which we've been placing in HP executives' beds for the past few months in hopes of getting our calls returned. No luck so far.

1
1
This topic is closed for new posts.

Forums