General Discussion

Highlighted

SSD Aggregate Best Practice

Hi NetApp Gurus,

 

We have recently bought new SSD disk shelves. I am looking best practices suggestion for the aggregate.

Following below are my questions. Looking forward for suggestions.

 

1) What is the NetApp best practices to provision aggregate so we can get good performance and make it scalable at the same   time.

2) What is the suggested good size for aggregate to get good performance. I understand that 64 bit aggregated can be made 100 TB in size. But making big aggregates can decrease the performance ?

3) New Disk shelves has 1.5 TB SSD disks and old disk shelve has 600 GB SSD disks. Can we mix 2 different size of disks in 1 aggregate  and still get good performance. Is mixing 2 different sizes disk suggested ?

4) For charging storage to customers what is the good idea, out of the 2 below .

     i) Make dedicated aggregate to each business unit and assign it to 1 SVM ?

     ii) Make dedicated aggregate and dedicate SVM for each business unit ?

 

1 REPLY 1
Highlighted

Re: SSD Aggregate Best Practice

Link to the aggr power guide,  this will help a lot.   

https://docs.netapp.com/ontap-9/topic/com.netapp.doc.dot-cm-psmg/Disk%20and%20aggregate%20management.pdf 

 

 

1) What is the NetApp best practices to provision aggregate so we can get good performance and make it scalable at the same  time.

- typically for most systems a 50/50 split across both controllers in an HA pair.   with SAS drives you have to be more so aware of spindle count per raid group in an aggr.    max aggr size can vary based on controller type/os version so check hwu.netapp.com for details.    

 

2) What is the suggested good size for aggregate to get good performance. I understand that 64 bit aggregated can be made 100 TB in size. But making big aggregates can decrease the performance ?

- no 

 

3) New Disk shelves has 1.5 TB SSD disks and old disk shelve has 600 GB SSD disks. Can we mix 2 different size of disks in 1 aggregate  and still get good performance. Is mixing 2 different sizes disk suggested ?

- mixing drives in an aggr can be tricky.    You would want to have seperate raid groups inside the aggr.   Also note that once disks are in an aggr, they cannot be removed.  So if you have drives that are End of support in 2021 and the other type of disk End of support in 2023.  you would be in an odd situation where part of the aggr might not be under support.    

I would also not mix same size drives of different speeds.  for example. the 3.8TB drives have a 6 and 12 gb sas speeds,  keep them in separate aggrs.   The best bet is to KISS it. (Keep it simple, stupid) 

 

4) For charging storage to customers what is the good idea, out of the 2 below .

     i) Make dedicated aggregate to each business unit and assign it to 1 SVM ?

     ii) Make dedicated aggregate and dedicate SVM for each business unit ?

 

don't dedicate an aggr to a biz unit. unless you have at max 2 business units...   The SVM is good logical isolation of workloads.   I would make each biz unit it's own SVM.   but splitting SAN/NAS workloads between SVMs.   example.  Marketing needs both FC and SMB.    Make separate  marketing_fc and marketing_smb svms.    not a requirment, but I find it easier to manage and track. 

Check out the KB!
Knowledge Base
All Community Forums