Are the HP LeftHand P4500 SAN's any good?

C

cyr0n_k0r

Supreme [H]ardness
Joined
Mar 30, 2001
Messages
5,360
I had a demo today of an HP P4500 Lefthand SAN. Looked pretty good. We are an all HP shop for our servers and I'd like to keep things in the product family. Just wondering if the LeftHands are solid units.

We are looking a their 14.4TB 2 node cluster. 24 LFF 600gb 15k rpm SAS.
This would replace our 2 existing MSA G1 and G2's. Add capacity too. We would ideally move all our VM storage off our hyper-v hosts onto the SAN so I could finally start doing some hyper-v clustering.

I figured I would post this here instead of the disk sub-forum since you guys might have more experience with SAN's. And it's mostly for our virtualization anyway.
 
Eh. They are a pretty standard iSCSI box. Nothing fancy. Should be cheap. I don't see a lot of it out there since HP doesn't push it as much. Equallogic is a better answer... EMC is better than both. ;)
 
Search here for Compellant. We had a good discussion on it not long ago.
 
Does anyone still like Pillar data systems? I thought some of their stuff is ok.
 
I have a Lefthand p4500 system at work. Jason hit the nail on the head, basic, is exactly what it is. There is no comparison to the VNX etc. Basically, it's a bunch of Proliant Servers, DL380's, with a 410 storage controller in each. Then they load a basic customized small footprint Unix/Linux OS which handles the storage clustering and is centrally managed by a small software package, Central Mangement Console. You do get basic replication and snapshots, though the replication is far from anything that you can get with the VNX. Each node is comprised of a RAID 5 though you can set it up for RAID10. For redundancy, you run a network raid level, basically node failure protection.

Now, I wouldn't say it's all that bad, you can easily expand..etc...however, for the money there is better.
 
Last edited:
Vader said:
I have a Lefthand p4500 system at work. Jason hit the nail on the head, basic, is exactly what it is. There is no comparison to the VNX etc. Basically, it's a bunch of Proliant Servers, DL380's, with a 410 storage controller in each. Then they load a basic customized small footprint Unix/Linux OS which handles the storage clustering and is centrally managed by a small software package, Central Mangement Console. You do get basic replication and snapshots, though the replication is far from anything that you can get with the VNX. Each node is comprised of a RAID 5 though you can set it up for RAID10. For redundancy, you run a network raid level, basically node failure protection.
Click to expand...
I've been quoted the 14.4TB 2 node starter for about $28k "re-new" or about $36k for brand new. Worth it?
 
You really have to think about it. I'm guessing you are getting two nodes, 600GB 15k sas? Usable space after all the over head is a bit over 5TB. Then you have to factor in the network RAID.
 
Vader said:
You really have to think about it. I'm guessing you are getting two nodes, 600GB 15k sas? Usable space after all the over head is a bit over 5TB. Then you have to factor in the network RAID.
Click to expand...
Uhh what?
how do you go from 14.4TB to 5?
I wouldn't be synchronously replicating every volume, so we wouldn't necessarily have to reduce the space by 50%
 
I didn't point out 5TB per node.....and yes..5.32TB useable PER NODE.



We run Network RAID10 for all PROD Volumes. For the DEV side we run as is. I will also point out that the support is questionable until you get to tier III.
 
Last edited:
It's a SAN, so why does it matter how much space is on each node? Isn't it all just one big lump of storage? Or do you have to worry about each node being managed independently.
 
It's all one big pool, however, i'm pointing out that the 14.4TB SAN is not useable capacity. You lose a lot before you even go network RAID. Assuming you did network RAID 10 on all your volumes, you would only have 5.32TB's of useable capacity or 10.64TB's standard Network RAID0.

In conclusion:

1. You lose per node w/hotspare.
2. You lose per node w/RAID5 or RAID10 (obviously you lose more with this option), up to you.
3. You lose again for network RAID if you go that route.

Whereas a standard SAN you would lose the hotspare but it's not per DAE like the Lefthand where each node is penalized a hotspare. Then you can create your Storage Pools or RAID groups so you get penalized there, but you do not have network RAID to factor in.

They claim you can tier but it's manual..so why bother. Like a lot of vendors they throw drives @ IOPs, which, to me, is crazy with Flash available, which they don't support, BTW. On the positive side, it's easy to manage, and install, and you are up and running very quickly...but again, look at what you are configuring, basically a bunch of servers with drives using the internal array all in a cluster.
 
Is the NetApp FS2000 series a better route?
I'm somewhat price sensative, so when my sales rep said about 28k for the P4500 2 node I was pretty happy.
 
NetApps fiscal year is ending in March and they had already-great pricing on the 2040 when I purchased one in December, so you could probly swing a sweet deal.
 
OP: They're solid devices. No issues, no problems, just work, and run well. Good VAAI support too, as well as SRM.

Vader said:
I didn't point out 5TB per node.....and yes..5.32TB useable PER NODE.



We run Network RAID10 for all PROD Volumes. For the DEV side we run as is. I will also point out that the support is questionable until you get to tier III.
Click to expand...

Support is questionable for EVERY san vendor till you get to tier III. :p
 
Let me give you an example of why I think there are better products out there.

We have a client that specifically requested HP, no problem we get the Lefthand SAN for their environment.

For another client I replaced a Clariion with a VNX5500.

Comparing the two, the original config at the time for the was 5 P4500 nodes at 26.6TB useable along with the HP Procurves iscsi network equipment and cabling. Keep in mind the loss of storage for network RAID, I actually had about 12TB's of useable capacity.

For the VNX5500 for a very similiar price I got:

1. VNX5500 ALL PROTOCOLS
2. 27.5TB
3. FAST CACHE/FAST VP
4. Local Protection Suite
5. Dedupe/compression/
6. Cisco 8Gb Fiber Network
7. All the IOPS I need.
8. Phenomenal Integration with VMware
9. Excellent managment suite with Unisphere.

In my situation, absolutely no comparison. Along with the points I made above, hands down there is better out there for the money.

On top of that..I had real issues and problems with the lefthand system. It took weeks of misinformation from Tier 1 and 2 support to only escalate through our corporate contacts to work with a Tier 3 engineer.
 
lopoetve said:
OP: They're solid devices. No issues, no problems, just work, and run well. Good VAAI support too, as well as SRM.
Click to expand...
The netapp or the Lefthand?
Vader said:
For the VNX5500 for a very similiar price I got:
8. Phenomenal Integration with VMware
9. Excellent managment suite with Unisphere.
Click to expand...
We don't have VMware. We have Hyper-V. So I need something with Phenomenal integration with Hyper-V. I'm sure the EMC could too, but just want to throw that out there.
 
cyr0n_k0r said:
The netapp or the Lefthand?
We don't have VMware. We have Hyper-V. So I need something with Phenomenal integration with Hyper-V. I'm sure the EMC could too, but just want to throw that out there.
Click to expand...

Either. Both are good solid basic arrays.
 
The issues with NetApp is that you have to have full disk trays and there is no mixing disks in the trays
 
I think one point no one has mentioned is the potential need for a Fail Over Manager on the Lefthand config- if you have an even nos of storage boxes. This is a right pain in the arse since you need to find somewhere to host this (so ESX / Hyper-B host or via VMware player)

Welcome to be told I am incorrect though. :)
 
haileris said:
I think one point no one has mentioned is the potential need for a Fail Over Manager on the Lefthand config- if you have an even nos of storage boxes. This is a right pain in the arse since you need to find somewhere to host this (so ESX / Hyper-B host or via VMware player)

Welcome to be told I am incorrect though. :)
Click to expand...

Toss it on a spare box. That's what I do - 3 LHN clusters in different geographical locations, each has a FOM for one of the other clusters.
 
lopoetve said:
Toss it on a spare box. That's what I do - 3 LHN clusters in different geographical locations, each has a FOM for one of the other clusters.
Click to expand...

Yeah I know thanks. I went through a reseller and they kind of missed this. I didn't have a spare box in that security zone so was quite annoying :) Also seems kind of archaic to me but maybe this is the norm in storage? (not really my SME forte I must admit!)
 
You must log in or register to reply here.
Back
Top