This post is more than 5 years old
5 Posts
0
28299
3 Dell Equallogic PS6010's
Hello,
I have 3 Dell Equallogic PS6010 Raid 6 Accelerated what is the optimal configuration for performance and reliability. Should i put all the members in the same pool with a few volumes?? using this configuration what if one of the members die in the pool will the volumes still be online?? Or should I have a separate pool for each member and syncrep accross?
Thanks!
Anonymous
5 Practitioner
5 Practitioner
•
274.2K Posts
1
September 30th, 2014 12:00
Hello Quackattack,
If you have multiple members in the same pool, then by default data for each volume is spread out across the members. Up to three by default, unless the volume is too large to fit on just three members.
So in the event of a member failure, all volumes in common with that member will go offline until that array is restored to service. Members in the same pool work together so that will provide the best performance but at a greater risk. The arrays are highly redundant to prevent that. Also, members in same pool with swap out blocks to keep the I/O latency in balance between the members. So a member with lots of I/O will send busy blocks to another member, and take in idle blocks from that member. Keeping latency in check and space balanced as well. That's the Advanced Performance Load Balancer.
But if that is a priority then putting them into separate pools would be better. You would then have to manage the IO loads manually to prevent one pool from being saturated, whiile the others might be idle by comparison.
You can also do a combo. Put two members in one pool, then another member in another to do syncrep on your most critical volumes. Giving you a balance between too.
Much depends on the load you are expecting to apply to the group. Also, if you can levarge EQL MPIO enhancements, like MEM for ESXi, HIT/LE for Linux or HIT/ME for Windows.
Also if you are using ESXi, VMware requires Enterprise or Enterprise+ to run MPIO extensions like MEM. If you can't use MEM, then this PDF has info on how to improve MPIO fuction using standard VMware Round Robin. Along with other tips as well.
en.community.dell.com/.../20434601
If you haven't done so already, install SANHQ v3.x to better monitor your group(s).
Regards,
quackattack
5 Posts
0
September 30th, 2014 12:00
Thank you so much for your help! So would you recommend this configuration 2 members with our fastest drives in 1 pool, 1 member in a separate pool with syncrep and vmware MPIO to all members providing redundancy and performance
Anonymous
5 Practitioner
5 Practitioner
•
274.2K Posts
0
September 30th, 2014 13:00
I can't recommend which option without knowing what your perf needs are projected to be. Writes on sync-rep'd volumes won't be ack'd until they reach the alternate pool. So if the member is not on same switches as primary members, then any additionaly latency there will also impact write performance.
You can also start with all three in same pool, then if you don't use up all the space initially, you can remove a member later and put it into its own pool. So you can use SANHQ to monitor it and decide if you have enough I/O capacity with just two members. Or even do reverse. Start with 2 + 1, then remove syncrep later and merge the member back into the pool with the other two. PS architecture is very flexible that way.
What are the specifics on the 3x arrays? What EQL FW are you running? What kind of switches are you using and what's the interconnect between the switches.
Re: MPIO. Regardless of how you configure the group, MPIO is always top recommendation. Install MEM if possible, especailly on multi-member pools.
quackattack
5 Posts
0
September 30th, 2014 14:00
They are 3 PS6010 8 SAS 3TB and 8 SSD drives EQL FW 6.0.7 the switches are dell powerconnect with fiber 8024F. One other thing i was curious about was how my 3TB drives only show as 750GB the model number is ST3000DM001
Anonymous
5 Practitioner
5 Practitioner
•
274.2K Posts
0
September 30th, 2014 14:00
OK, so this isn't a Dell provided configuration. Since Dell never mixed SATA 3TB and SSD drives and especially not on the 6010 model. And no hybrid model was ever 8+8.
I suspect they show up at 750 because the controller wasn't quallified with 3TB SATA drives. The largest SATA drive was 2TB, only 3TB drive supported was a Near Line (NL) SAS 3TB drive.
Given this is not a supportable configuration, I would not put this into production. No way to know what problems you could have with those drives, in that configuration especially. I would not risk my data on such an unqualified storage device. Too risky in my opinion.
Anonymous
5 Practitioner
5 Practitioner
•
274.2K Posts
0
October 6th, 2014 13:00
Hello quackattack,
I've been thinking about your set up and I wanted to add that I suspect, but can't prove, that the 750GB vs. 3TB is because the capacity has rolled over. If so, that would increase the risk of data loss or corruption.
Anonymous
5 Practitioner
5 Practitioner
•
274.2K Posts
0
October 10th, 2014 13:00
Hi Quackattack,
The quick answer is put them in the same pool. That way they work together and provide best performance and easier management. If you put them in pools, then you might have to move VMs or Datstores around to balance out the I/O.
However, should you ever have a total failure on one member, then all the volumes would be offline until you fixed that member. So for some customers this isn't acceptable. Or do some combination, one pool for critical VMs, and then two members in another pool for most VMs.
So it's what is best for you.
I would still upgrade to at least 6.0.11, or 7.0.9 to ge the latest fixes and features.
Regards,
quackattack
5 Posts
0
October 10th, 2014 13:00
My ideal setup was as stated in my previous post to get more space but its sounding like I should just stay with my current setup 8 x 600GB 15K SAS 8 x 100GB SSD HDD RAID 6 Accelerated.. I have three PS6010's with this config. What is the recommended setup for a standard virtual environment??
Thanks Donald!
QuackAttack