number of discs relative to iops

[Aleskey_K]

Good afternoon, Colleagues! Please explain whether there will be a large increase in the performance of storage systems when using a system with 30 disks, in comparison with a system with 60 disks. The disks in the system are the same and the raw volume is not important. Thank you in advance

[Ontapforrum · Accepted solution]

In general, more disks means more parallelism, more IOPS. However, this is not applicable to 'Optimised Specialised Storage Systems'. These days most of the newer NetApp systems do 'abstracted IOPS' (Running :cDOT/ONTAP). The total IOPS that a node can provide is based on the physical characteristics of the node—for example, the number of CPUs, the CPU speed, and the amount of RAM. The total IOPS that an aggregate can provide is based on the physical properties of the disks—for example, a SATA, SAS, or SSD disk. Hence, the combination, CPU cores, RAM in your Node, and the number of volumes on that 'flash-pool, IOPS availability will differ but will be way greater than the actual RAW IOPS. It's bit complex but I guess its projected based on a generic workload based testing and it's different for different filer Model.

Regarding your specific query: More disks you have in your raid-group set, more IOPS will be available at RAW aggregate level (That doesn't mean putting all in the single-raid_group, there is always trade-off). As I mentioned, the actual front-end IOPS will depend on many different factors.

Considerations for sizing RAID groups:
https://docs.netapp.com/ontap-9/index.jsp?topic=%2Fcom.netapp.doc.dot-cm-psmg%2FGUID-EE30CDA9-095F-4DD1-9A29-BB9F0E55DB4F.html&anchor=:~:text=When%20y....

[Aleskey_K]

Thanks for the quick reply, very detailed and very clear.
I want to clarify about concurrency. If we take the entire set of disks, regardless of the use of the raid level (6,10,50,60), what algorithm is used to write to the RAW array? Does recording happen on all discs simultaneously or does ONTAP select the fastest discs and write to the first one very much on them?

[Ontapforrum]

Yes, it stripes across all disks, but "how it does" is perhaps already covered in tons of articles on the NetApp knowledge base/Community-forums as well as on the various blogs on the Net.  The key word here is 'WAFL', just explore for this info.

In short : NetApp storage systems comes "write-optimized" through the use of write cache and nonvolatile memory (NVRAM or NVMEM) along with the core component called 'WAFL'. Following articles will give you some useful idea on this topic.

Articles:
https://kb.netapp.com/Advice_and_Troubleshooting/Data_Storage_Software/ONTAP_OS/How_does_WAFL_and_striping_distribute_data_among_disks%3F
https://kb.netapp.com/Advice_and_Troubleshooting/Data_Storage_Software/ONTAP_OS/FAQ%3A_Consistency_Point
https://community.netapp.com/t5/Tech-OnTap-Articles/Back-to-Basics-RAID-DP/ta-p/86123
https://docs.netapp.com/ontap-9/index.jsp?topic=%2Fcom.netapp.doc.dot-cm-psmg%2FGUID-2808F15F-6057-4E43-9407-5B3CA274E6C5.html