VMware Cloud Community
hutchingsp
Enthusiast
Enthusiast

Equallogic Feedback (and vs. Lefthand)

I've been looking at solutions for putting in up to 5tb of storage for a project we have coming up, with a view that whatever we put in now may well be utilized when our main SAN/ESX platform is replaced in around a year.

Lefthand's VSA appliance is one potential option, cheap enough, flexible enough, but obviously DIY even with support on all the hardware and software.

Today I received a very good quote from Dell on an 8tb SATA Equallogic PS4000E dual controller.

At a basic level the two solutions seem similar i.e. all licenses bundled so no extras/shocks down the line, "node" expansion model i.e. need more space/resilience buy more nodes and disperse them.

What I'm unclear on is the specifics that separate the two, and what Equallogic is like from a management/admin/flexibility viewpoint (I downloaded the Lefthand VSA demo so tried for myself).

For the iminent project the thing would hold LUNs containing SQL databases assigned to a VM, most likely direct to the OS using MS iSCSI initiator rather than as VMFS/VMDK. For this project the database may get big, but transactions/throughput is very low. If and when we replaced our main SAN we're looking at around 30 VM's, low usage mainly "One VM one application" with a virtualized file server (currently serving 8tb via RDM) and a virtualized Exchange server, which by that time should be on Exchange 2010 so reduced I/O requirements.

I'd appreciate any feedback particularly on the Equallogic and what it brings to the table.

Reply
0 Kudos
12 Replies
Chuck8773
Hot Shot
Hot Shot

We have used EqualLogic for five years now. I like their system and it is a nice fit for smaller deployments. One issue we ran into was the use of SATA disks for the VMFS volumes that store the system vmdk's. During a boot storm, like a massive patch process, VM's were creeping upwards of 30 minutes to boot. Windows really does not like that. Many services failed to start. We switched to SAS disk and now, during boot storms, VM's take around 2-5 minutes to boot. SATA is still fine for the data volumes directly connected from within the VM. With EqualLogic, one array = one disk type. Some other vendors let you put multiple types of disk in the same array.

Overall, EqualLogic - Solid product, easy to manage, great support, limitations when you get beyond about 15 hosts.

Charles Killmer, VCP4

If you found this or other information useful, please consider awarding points for "Correct" or "Helpful".

Charles Killmer, VCP4 If you found this or other information useful, please consider awarding points for "Correct" or "Helpful".
Reply
0 Kudos
hutchingsp
Enthusiast
Enthusiast

Thank you for the reply, appreciate it.

I take the point about SATA vs. SAS, and to be clear, for this project I don't see SATA being an issue, though if we were to put more EQL in when our main SAN is replaced I'd almost certainly be looking at SAS as that would be expected to run all of our production VMs. At that point, if we did buy a PS4000 now I'd be looking to perhaps use that at our DR site, or as a different tier (haven't got my had around how EQL handle different types of storage yet) for "dum" file server data.

Our Dell rep was saying that even she has set one up and used one on a demo day, which isn't such a huge deal to me (not being big-headed in saying that, but it's my job to do this stuff) but I could see that being a bonus should there be an issue and I'm not around.

Do you have any experience of the replication? As I mentioned I'm also looking at Lefthand and they do realtime "Network RAID" between units - I'm unclear right now if you have two EQLs at opposite ends of your site if they replicate in real time or at intervals and if it's "A to B" replication or if the two units work as a cluster with redundancy - I don't imagine they do as if you have a fast SAS unit at your main site and a slower SATA unit at your DR site the performance split would be uneven.

Reply
0 Kudos
Peter_Grant
Enthusiast
Enthusiast

I'll start by saying I haven't used Equallogic so I may be biased however LeftHand is a good solid solution and suitable for Enterprise deployments, it has all the big SAN features. Replication gives the ability to lose 1 or more nodes but you lose half the storage space. However the more nodes you have the more IOps you get.

I guess my point is LeftHand works fine and is very easy to manage in house.. Not sure if that's much help! Smiley Happy






----


Peter Grant

Virtualization Consultant

Xtravirt.com

Please award points if you found this helpful

------------------------------------------------------------------------------------------------------------------- Peter Grant CTO Xtravirt.com
Reply
0 Kudos
hutchingsp
Enthusiast
Enthusiast

Thanks Peter, it helps and I've downloaded and been trying the VSA demo to get a feel for the technology - looks nice, but I think out of our price range (obviously I can't divulge prices but Dell to like to offer us good pricing), and as much as I like the Lefthand kit, I do have concerns about support under a global giant such as HP vs. when they were a smaller independent.

Back on the EQL front, I have some webinars to watch on their website that may make it all clear, but any info on how the clustering/replication works would be grand.

Reply
0 Kudos
Peter_Grant
Enthusiast
Enthusiast

Price is always a big factor and Id be interested to know if its a big difference between the two.

Make sure when you're buying to find out how much USEABLE storage you actually get. With Leftand when you enable 2-way replication you can lose one node and stay online but you also lose 50% of your raw storage. This is the best case scenario. You can enable 3 or 4 way replication but then you only get 33% or 25% usable storage. Most people use 2-way for one node lose and 50% storage lose. Be aware that buying '1' LeftHand SAN actually consists of 2-nodes so you have the failover option straight off.

Replication is real time and requires Gigabit and something like 2ms or 5ms latency...basically a Gigabit LAN.

There is also Remote copy which is async replication which can be across a remote WAN link and supports SRM. Also one good thing about lefthand is that the snapshots are thin, so if you're snapshoting or doing remote copy on a 1TB volume only using 100GB the snapshot will also reflect this 100GB.

Again key point is to work out usable storage. I've got some LeftHand sizing spreadsheets, if you tell me the number of GB / TB usable you have and your IOPS requirement if you have it then I can tell you what I think you should use.

I'm not a LeftHand rep, I've just done a bit of sizing on my last project so it's fresh in my mind.






----


Peter Grant

Virtualization Consultant

Xtravirt.com

Please award points if you found this helpful

------------------------------------------------------------------------------------------------------------------- Peter Grant CTO Xtravirt.com
Reply
0 Kudos
hutchingsp
Enthusiast
Enthusiast

For the project I'm looking at now, we need approx 5tb usable space, right now we don't need replication but things like writeable SQL snapshots would be useful as the data being stored is predominantly low transaction but high total volume databases.

When we replace our primary SAN, I would like to use replication as we have a large site with fiber everywhere so it makes sense to replicate to our DR room.

Problem is the money I'm likely to have available now is for this project, not for the future requirement, so essentially I'm trying to put in a standalone foundation now that can be expanded in a year or so.

With Equallogic this seems simple, buy one unit now, buy another later.

With Lefthand it's simple with, say, the VSA, which I've tried using our existing ESX environment, and which I like a lot, but with their hardware solution you're buying two nodes by default so admittedly you get all the replication capability from day-one, but you're paying for it from day-one too, which is where it would be a problem financially as the current project doesn't justify it.

Essentially right now if I were looking at Lefthand for this, I think it would be only at the VSA, not the hardware product, and Dells pricing doesn't appear to cost much more than the VSA.

Reply
0 Kudos
Peter_Grant
Enthusiast
Enthusiast

Here's one option based on 5TB requirement.

Default 2-way configuration (you can lose one node and SAN stays online)

P4300 SATA Starter SAN, Qty 1 (2-nodes), Usable Storage 4.58 TB, 800 IOPS, SATA Disks

If you buy just one expansion SATA SAN node then you'll get the same capacity (half the price) with less IOPs and no failover. You'll need to speak to LeftHand tech sales but maybe this would do for a small environment. as an alternative to VSAs. We have put VSAs in before for small VDI environments of about 20 users and works ok. As a ROUGH rule of thumb if you don't know your IOPS requirement you could use 50 IOPS per server VM as a very general rule of thumb to see if you're in the ball park.

I'm not trying to say EL isn't a better solution for you. It may be better to pay slightly more and get a hardware EQ solution rather than VSAs. Remember with VSAs you still need to provide the physical storage.

Peter






----


Peter Grant

Virtualization Consultant

Xtravirt.com

Please award points if you found this helpful

------------------------------------------------------------------------------------------------------------------- Peter Grant CTO Xtravirt.com
Reply
0 Kudos
hutchingsp
Enthusiast
Enthusiast

In an ideal world I'd much sooner be looking at the physical appliance vs. VSA, and I think once I know the way the project is going i.e. timescales, I'll get some real world pricing on the Lefthand.

How have you found Lefthand support? It's one area that concerns me given the HP takeover.

I've watched a couple of Webex's on the Equallogic and how their provisioning and snapshots and replication works and it looks good, similar in many ways to Lefthand but there are some fundamental differences such as that any spanned volumes with EQL appear to break if you lose a node, in fact that seems the single biggest downside with EQL - that the redundancy is only within each frame, lose a frame and you lose access to whatever was stored on that frame as that's the only place it's stored (unless you have it replicated in which case you're into manual failover AIUI?).

Reply
0 Kudos
Peter_Grant
Enthusiast
Enthusiast

Spanning volumes in LH is never recommended unless 2-way replication is enabled. They sell the appliances in pairs for just this reason. If u use replication then the nodes form a cluster and volumes are spread over them. I assumed EQ would offer the one node failure option if not then that's a big difference. Never contacted LH post sales support but my pre sales tech support has been good.





----


Peter Grant

Virtualization Consultant

Xtravirt.com

Please award points if you found this helpful

------------------------------------------------------------------------------------------------------------------- Peter Grant CTO Xtravirt.com
Reply
0 Kudos
hutchingsp
Enthusiast
Enthusiast

I didn't know you could span in LH without replication actually, I figured "no replication" meant the enire LUN stayed on a single node unless I've misunderstood what you meant?

Be interested if anyone can clarify on the EQL stuff (I can obviously ask Dell) but my understanding was that the resilience is via redundancy on the individual boxes, the actual cluster isn't redundant.

Reply
0 Kudos
Peter_Grant
Enthusiast
Enthusiast

I'm stating this from memory so could be wrong but I think u can span a volume. With LH u creat a cluster with one or more nodes in it. When u create a volume it's a volume within the cluster so it should be spread amount the nodes. Could be wrong but pretty sure that's how it works. Choosing no replication , 2-way etc only specifys the network RAID level hence redundancy





----


Peter Grant

Virtualization Consultant

Xtravirt.com

Please award points if you found this helpful

------------------------------------------------------------------------------------------------------------------- Peter Grant CTO Xtravirt.com
Reply
0 Kudos
ericsl
Enthusiast
Enthusiast

hutchingsp,

Not sure if you made a decision yet but I can offer you my experience with the Dell Equallogic series.

1) It's inexpensive. This may not be your top issue but it's usually right on up there for most people...

2) Equallogic tech support is top-notch. I've had to call in several times with a variety of questions and always got prompt profeesional help. In my mind this is the most important criteria (after price Smiley Happy)

3) Replication works great, but it is only one volume to one replication SAN, not one-to-many. Where this rears it's head is in the next issue.

4) No Synchronous replication. Period. If your SAN group member (any member) has a backplane failure your are down. That's all there is to it. There is no Active/Active scenario is the Equallogic world between groups. Now comes in point number 3, this means that you can not replicate both locally and remotely. You have to choose. In my case we choose to user another solution for either the local or remote (depending on the server) replication of VMware virtual windows servers.

5) Supports VMware SRS. You probably won't care about this in a lab but if you transition this box to production SRM makes it much easier to failover to another SAN. The caveat is that you have to choose whether that SAN is local or remote. You can't have both. See point 3...

6) Performance has been awesome for us. I know that "awesome" is a subjective term but we are running around 70 VMware virtual machines and 6 Solaris hosts on the a PS6500 and it is coasting along. A pool of these would be very fast.

I could go on here but I am very satisfid with the PS series, except for the lack of one-to-many or Group Synchronous Replication.

Eric Langley

myManagedBackup.com

Reply
0 Kudos