Anyone have experience with an MD3000 and configuring it for maximum performance?
I'm about to purchase an MD3000 and I'm looking for some feed back on the good the bad and the ugly.
I've read the Array tuning document that they have for the MD3000 and MD3000i. A lot of good info in there. Does anyone know if the active/active arrays do load balancing? I want to try and maximize the use of both RAID controllers. I'm wondering how well these work compared to a perc5e or 6e controller. If anyone has any experience with this I would love to get some feedback. The spec sheet identifies 512MB of cache. Is that enough for an enterprise level product?
Also, I'm trying to figure out which disk configuration will give me the best performance while trying maximize capacity. I want to try and keep cost down while still having a good data protection scheme. For a fully loaded array (15 Disks) I'm debating between the following configurations.
RAID 10 using 14 1TB SAS 7.2k RPM w/ HS
or
RAID 5 using 7 450GB SAS 15k RPM w/ HS
RAID 5 using 6 450GB SAS 15k RPM w/ HS
With the first option, I'm getting almost 2TB more and I'm hoping that performance won't be worse than using the RAID5 option. In addition, it's a less expensive solution.
Any storage experts with some thoughts on this?
I've read the Array tuning document that they have for the MD3000 and MD3000i. A lot of good info in there. Does anyone know if the active/active arrays do load balancing? I want to try and maximize the use of both RAID controllers. I'm wondering how well these work compared to a perc5e or 6e controller. If anyone has any experience with this I would love to get some feedback. The spec sheet identifies 512MB of cache. Is that enough for an enterprise level product?
Also, I'm trying to figure out which disk configuration will give me the best performance while trying maximize capacity. I want to try and keep cost down while still having a good data protection scheme. For a fully loaded array (15 Disks) I'm debating between the following configurations.
RAID 10 using 14 1TB SAS 7.2k RPM w/ HS
or
RAID 5 using 7 450GB SAS 15k RPM w/ HS
RAID 5 using 6 450GB SAS 15k RPM w/ HS
With the first option, I'm getting almost 2TB more and I'm hoping that performance won't be worse than using the RAID5 option. In addition, it's a less expensive solution.
Any storage experts with some thoughts on this?
ASKER
Thanks... Interesting Read. If anything, I think it helps solidify my comfort level with going RAID 10 for my specific application, although, I still think that RAID 5 with a hot spare could still be viable in some situations. Especially where capacity is king and write performance isn't critical.
For the article on RAID 5.... Wouldn't a periodic checkdisk help prevent against a URE on reconstruction?
For RAID 10... are there any downfalls to striping against more disks (in my case 14)? With that configuration we're talking a 6.5TB volume.
For the article on RAID 5.... Wouldn't a periodic checkdisk help prevent against a URE on reconstruction?
For RAID 10... are there any downfalls to striping against more disks (in my case 14)? With that configuration we're talking a 6.5TB volume.
I still use RAID 5 with hot spare unless I have a client that want to take no chances.
yes check disk would help prevent against a URE and i would also recommend enterprise hard drive.
they have a better lifespan and CRC check.
To this day i have yet had one fail. I generally run out of space first
http://www.tomshardware.com/charts/enterprise-hard-drive-charts/benchmarks,27.html
CT
yes check disk would help prevent against a URE and i would also recommend enterprise hard drive.
they have a better lifespan and CRC check.
To this day i have yet had one fail. I generally run out of space first
http://www.tomshardware.com/charts/enterprise-hard-drive-charts/benchmarks,27.html
CT
ASKER
Any thoughts on the MD3000 controller?
Why did you choose these cards. Are you running Dell Server Hardware?
Forced Airflow is Required
The Tj maximum temperature is 110C. However, do NOT run the IOP333 passively. The heatsink needs force airflow. Intel's thermal analysis used a heatsink of the same size but with more fins (hence better). However, they require a minimum of 200LFM with there heatsink. Over the surface the PERC stock heatsink, that is at least 4CFM. Do realize that if you used a 80mm fan, you would need a higher CFM rating of around 16CFM. This is assuming that the 80mm fan is next to the HS.
Bottom Line: Make sure to force air cool the PERC 5/i CPU. The card is designed for Dell servers with forced air.
I found this at http://www.overclock.net/hard-drives-storage/359025-perc-5-i-raid-card-tips.html
I personally like http://www.pc-pitstop.com/sas_controllers/adp2045.asp
CT
Forced Airflow is Required
The Tj maximum temperature is 110C. However, do NOT run the IOP333 passively. The heatsink needs force airflow. Intel's thermal analysis used a heatsink of the same size but with more fins (hence better). However, they require a minimum of 200LFM with there heatsink. Over the surface the PERC stock heatsink, that is at least 4CFM. Do realize that if you used a 80mm fan, you would need a higher CFM rating of around 16CFM. This is assuming that the 80mm fan is next to the HS.
Bottom Line: Make sure to force air cool the PERC 5/i CPU. The card is designed for Dell servers with forced air.
I found this at http://www.overclock.net/hard-drives-storage/359025-perc-5-i-raid-card-tips.html
I personally like http://www.pc-pitstop.com/sas_controllers/adp2045.asp
CT
ASKER
I am plan on using all dell hardware. 2950 server connected via a SAS HBA to MD3000. The RAID controllers reside within the MD3000. I didn't choose the cards. The hardware ships with it standard.
Not too worried about airflow in this case.
I'm more interested in knowing if anyone has had good/bad experiences with Dell's MD3000 DAS. Specifically, I want to know if they do a good job with load balancing and failover and overall performance.
Not too worried about airflow in this case.
I'm more interested in knowing if anyone has had good/bad experiences with Dell's MD3000 DAS. Specifically, I want to know if they do a good job with load balancing and failover and overall performance.
Ah as for performace i have been to several clients that have not compain about there unit.
CT
CT
ASKER
I'll be doing some performance testing on the unit later this week. I'll be testing RAID 10 vs RAID 5. It will be interesting to see if the controller can keep up with the drives.
ASKER
I completed the performance testing. Based on my results... The biggest performance difference was on the write test. It was quite a bit slower with RAID5.
I personally go with raid 10 when I can. Raid 5 is ok but not as much data protection as Raid 10. if two drives fail at same time data lost.
CT
CT
ASKER
Yes but... if you have a couple of hot spairs in a RAID5 config, you can withstand 2 simultaneous failures. Is that right?
ASKER
Thanks for the article. I should have clarified a bit more. I plan on having 2 disk groups. Each with a hot spare. So... in this case I should be able to withstand 2 simultaneous disk failures as long as they are in different disk groups. If a double disk failure occurs within a disk group... then I'm out of luck.
Thanks
Thanks
ASKER CERTIFIED SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
ASKER
Thanks for your input.
http://www.cyberciti.biz/tips/raid5-vs-raid-10-safety-performance.html
CT