HPE EVA Storage
1751897 Members
5848 Online
108783 Solutions
New Discussion юеВ

MSA 1500 vs. EVA 5000 SAN

 
SOLVED
Go to solution
Jorge Cocomess
Super Advisor

MSA 1500 vs. EVA 5000 SAN

Greetings,

We are about to purchase (end of life) EVA 5000 with 8TB raw disk space for a very reasonable price. Even though the EVA 5000 is end of life, but support will still exist for another 3-5 years as I been told by HP. However, we're facing performance issue with the MSA 1500 SAN evironment. From what I've read, the EVA SAN is the way to go. Does anyone here know where I can find some comparision docs on the MSA stack against the EVA?? I just need some facts that the EVA will provide the double or triple what MSA SAN can give you.

I need to make the management team feel more warm & cozy.

Thank you in advance.
Jorge
36 REPLIES 36
Bernhard Mueller
Honored Contributor
Solution

Re: MSA 1500 vs. EVA 5000 SAN

Jorge,

look at HP web for the "Quick Specs"

high level numbers are

MSA1500 max 512MB cache
up to 30000 IOPs, max 200MB/s throughut

EVA5 2GB cache
up to 162000 IOPs, max 335MB/s throughut

Regards,
Bernhard
Jorge Cocomess
Super Advisor

Re: MSA 1500 vs. EVA 5000 SAN

Bernhard -- Thanks for the specs. Most of the EVAs comes with 2 controllers, right? What if you have 6 MSAs (RAID10, 128K chunk size) 512MB total with 4 luns per MSA, and you're seeing performance hit? Will 6 MSAs equal to 1 EVA or I will see the performance hit on these EVAs as well??

Regards,
Jorge
Mark Poeschl_2
Honored Contributor

Re: MSA 1500 vs. EVA 5000 SAN

Jorge -

How many servers and how many HBAs/server do you have? If you're percieving a 'performance issue' in the MSA configuration you describe, I'd suspect something on the server end first.
Paul Henderson_2
Frequent Advisor

Re: MSA 1500 vs. EVA 5000 SAN

If you have 6 MSAs serving only 4 LUNs per MSA under RAID 1/0, you could assume that there is some other issue within your SAN or on your hosts that is causing the performance degradation.

I'd work from your hosts to their HBAs, to the SAN switches to the MSA controllers to try to find the bottleneck. Sounds like you are experiencing numerous retries, path conflicts, zoning conflicts, or physical hardware issues.

That said, the EVA is a sterling piece of hardware, so you would not be disappointed to have one. But I'd wager it is throwing hardware at what is probably not the problem.
Jorge Cocomess
Super Advisor

Re: MSA 1500 vs. EVA 5000 SAN

5 servers and each server has one (1) HBA. Interesting point about zoning. As of now, all these servers are running RHAS 3.0 -- The profile on these MSAs are set to profile=linux

There are no other OSes on this Fiber switch. I need to look more into zoning per server. and per lun.

Thanks,
Jorge
Paul Henderson_2
Frequent Advisor

Re: MSA 1500 vs. EVA 5000 SAN

Jorge... several of my servers also run linux (redhat on some Dells and suse on a monster SGI), and all have only single HBAs. We did see performance and compatibility issues with the default linux drivers.

Simply put, the default qlogic hba driver (qla2300, in our case) did not properly support the storage array. We had to back out the default driver, and replace it with the HP supplied driver from their linux platform kit.

My notes, albeit cryptic and probably imcomplete are these:

# lspci
... snip
01:06.0 Fibre Channel: QLogic Corp. QLA2312 Fibre Channel Adapter (rev 02)
... snip

/* this shows you the qlogic hba */

/* now install the hp supplied linux platform kit. This should delete the current driver and replace it with the HP driver */

# cd /opt/hp/hp_fibreutils

/* you'll see a bunch of useful utilities that can verify your installation and also probe for attached luns */



John Kufrovich
Honored Contributor

Re: MSA 1500 vs. EVA 5000 SAN

Jorge,

As much I would like to see you spend the money.

If you like, I can take a quick look at your system. You will need to capture some data for me. Can you attach a cli cable to the 1500 and issue the following cmds.

>start perf

This will start the performance counters on the MSA controller. There will be very minimal controller performance impact.
Let this run during the busiest part of the day.

Then issue the following commands,
>show tech_support
>show perf
>show perf logical
>show perf physical

Cycle through the following commands a couple of times. Each take a snapshot at that moment.
>show taskstats
>show cacheinfo


>stop perf

Then attach the results.

Regards,

Jorge Cocomess
Super Advisor

Re: MSA 1500 vs. EVA 5000 SAN

Hi Paul,

here's the result after I ran "lspci" command;

40:1e.0 Host bridge: Hewlett-Packard Company zx1 Local Bus Adapter (rev 32)
80:02.0 Fibre Channel: QLogic Corp. QLA2312 Fibre Channel Adapter (rev 03)
80:02.1 Fibre Channel: QLogic Corp. QLA2312 Fibre Channel Adapter (rev 03)
80:1e.0 Host bridge: Hewlett-Packard Company zx1 Local Bus Adapter (rev 32)
a0:1e.0 Host bridge: Hewlett-Packard Company zx1 Local Bus Adapter (rev 32)

As you can see, the Qlogic (QLA2312) default driver was used on this server. Do we have something here?

Thanks for your time.

Jorge
Jorge Cocomess
Super Advisor

Re: MSA 1500 vs. EVA 5000 SAN

Hi John,

I will capture the info of my MSA 1500 sometime tomorrow and post it here for you to review. I really appreciate your help.

Thank you in advance.

Jorge