Unsolved
This post is more than 5 years old
6 Posts
0
1633
October 12th, 2010 14:00
Adding disks to ns120 performance question
We have one ns120 with 26 x 146gb 15k disks over two shelf’s which leaves us 4 free slots. Disk configuration is Raid 5, 4 plus 1, disk type is CLSTD
We wish to purchase more capacity the ns120 is solely being used purely for data storage via cif servers with no data bases or apps, options which emc supplied are below:-
Disk Type |
Usable Capacity per RG (R5 4+1) |
Total usable Capacity |
146GB (15k) |
534.5GB |
1069.1GB |
300GB (15k) |
1073.4GB |
2146.9GB |
Ideally I would like to go for 300gb option as this is more capacity, will this have lower performance then the 146gb raid group and can these be mixed and matched on the ns120 i.e. 146gb RG and 300gb RG.
Thank you in advance for your help.


dynamox
11 Legend
•
20.4K Posts
•
87.4K Points
0
October 12th, 2010 14:00
how much capacity were you planning on buying ? As you probably know EMC Celerra is very particular to raid group design, certain raid group configuration while valid for regular open systems will not work with Celerra, it will simply not recognize them.
TTown1
6 Posts
0
October 13th, 2010 00:00
At the moment another TB would be sufficient but if we can get more capcity without decrading performance i would rather go for the 300gb disks options and get an addtional TB of space even if its at an addtional cost.
thanks
dynamox
11 Legend
•
20.4K Posts
•
87.4K Points
0
October 13th, 2010 04:00
is it DAE 0_0 that has four empty slots ..at the very end right ?
TTown1
6 Posts
0
October 13th, 2010 04:00
yes correct, second shelf 4 slots right at the end.
cheers
dynamox
11 Legend
•
20.4K Posts
•
87.4K Points
0
October 13th, 2010 04:00
anther question, why 300G FC drives ? You can get larger FC drives these days, i have 450G in my NS80.
so you have a couple of options
1) let's say you go with FC drives (300,450 ..whatever). You can buy 6 drives. One drive will become your new hot spare and go into one of the empty slots in DAE 0_0. The other 5 drives will go into new DAE and configured in R54+1 configuration. This will ensure that when the space is discovered by Celerra, it will go into clar_r5_performance pool. Later one if you need more capacity, you can buy in 5 drive increments and keep populating that same DAE.
2) let's say you still go with FC drives. You can look at using 4+1 and 8+1 combinations (see picture). The problem with that is that 8+1 will end up in a different pool in Celerra. It's not a problem for new file systems but for existing file systems if you want to expand them in the future, you don't want them to span multiple pools (clar_r5_performance and clar_r5_economy)
3) How about SATA drives ? If this is truly for regular CIFS shares, SATA might be great for that. I have 150 terabytes of SATA in my NS80 serving thousands of concurrent users (students, staff, faculty). You can look in Celerra Monitor and look at disk performance during peak hours ..are you pushing your FC drives to the limit ?
umichklewis_ac7b91
300 Posts
0
October 13th, 2010 05:00
You can right-click on the storage array (in my case, a CLARiiON) and get a variety of disk stats.
TTown1
6 Posts
0
October 13th, 2010 05:00
Thanks for your swift reply, at the moment our 146gb drives are FC so we will be going for the 300gb FC drives, personelly i think we went overkill on the spec. I think the option is to purchase 11 new 300gb disks, configure these as 2x R5 (4+1) RG’s and have one as a hot spare as the 146gb hot spare will not work with this RG?Correct me if im wrong?
Do you think this option is advisable? Will the celerra see this in the same r5 performance pool? And lastly (sorry for all the questions) does the 146gb disk have the IOPS capbilty as the 300gb disks?
Also lol where do i find "disk performance" in "Celerra Monitor"
Once again thanks for you help with this its much appreciated
Thanks
dynamox
11 Legend
•
20.4K Posts
•
87.4K Points
0
October 13th, 2010 06:00
11 drives will work and yes ..you do need a 300G spare in that case
if you continue with this setup FC_RAID5_4+1_4+1_4+1 ..then yes, all LUNs will end up in the same r5 performance pool. 10k FC drives have ~140 IOPS, while 15k FC drives have ~180 IOPS. Obviously the 300G drive will have a lot more data on it and if all that data is actively accessed, then performance will be lower than a fully used and active 146G drive. For CIFS operations though ..i doubt you are reaching those IOPS.
TTown1
6 Posts
0
October 13th, 2010 07:00
excellent thank you, could you suggest some good reading material for navi manager? As i am a newbie to this, dont worry i will be booking the course sortly!
Thanks
dynamox
11 Legend
•
20.4K Posts
•
87.4K Points
0
October 13th, 2010 08:00
all Celerra related documentation can be found here:
Home > Support > Technical Documentation and Advisories > Hardware/Platforms Documentation > Celerra Network Server
good disk layout document
http://corpusweb130.corp.emc.com/ns/common/post_install/Configure_storage_Non_FC.pdf
kelleg
6 Operator
•
4.5K Posts
0
October 13th, 2010 09:00
Also, please rememeber to mark this discussion as "Answered" if you feel that your questionshave been answered>
glen
TTown1
6 Posts
0
October 14th, 2010 09:00
ok no problem
cheers