Expected Single User Write Performance FAS8200 HA Multi-Path

[Jmsimon]

We have an FAS 8200 HA Multi-Path system. It is connected to four disk shelves (two DS212 and two DS460c) via 12Gb SAS. It has four 10Gb connections to our data network(two LACP pairs). I'm trying to get an idea of the baseline write speed from a single Windows or Linux user to a single volume that is shared via CIFs or NFS. Currently, the max write speeds I have been able to achieve are around 3Gbps, while max read speed values have been as high as 6.5 -7 Gbps. I just want to make sure I'm not expecting the system to do something it isn't capable of, so does anyone know: is it possible to achieve at or near 10Gbps write speed to such a system over a CIFs or NFS network share?

[cedric_renauld]

HEllo,

The perfomance depend of many factors 😊

Onr of, is the number of axes, but you write you have many shelves, and I thinkls many disk, no issues.

another performance factor is the network, remember also you are in LACP, only one 10G can be used by a session, NFS or CIFS.

after you hace the protocol factor, NFS or CIFS, you can maybe ahve more options with NFS.

The read perfomance "6.5 -7 Gbps" seam food, but you can try to improve your wrtite performance.

For NFS if is possible try the pNFS, nut witrh 4.1 version :

https://www.netapp.com/media/19761-tr-4063.pdf

You can try to tuneup your NFS :

https://www.netapp.com/media/10720-tr-4067.pdf

Last option for a single user is to improve network card : 40 or 100 G 😁

But remember the Ontap system is made to serve large multiple users, one user isn't the good use case

[paul_stejskal]

https://kb.netapp.com/on-prem/ontap/Perf/Perf-KBs/How_to_engage_the_account_team_to_help_address_a_system_bottleneck
https://kb.netapp.com/on-prem/ontap/Perf/Perf-KBs/How_many_IOPs_can_a_FAS_or_AFF_controller_handle

[Jmsimon]

We'll reach out to the support team. In monitoring the Netapp's cpu, network, and disk usage during our test write operations we have been unable to detect any obvious bottleneck. Resource utilization on the Netapp seems to be hitting 25 to 30% at most. So my next thought was our network, but performing server to server file transfers over the same network using CIFs comes much closer to fully saturating the servers' 10 Gb network links.

[paul_stejskal]

https://kb.netapp.com/on-prem/ontap/Perf/Perf-KBs/ONTAP_9_Performance_Resolution_Guide Honestly this KB discusses all the steps to troubleshoot. It probably may be a network limitation.

[chenguanghui]

hello:

This depends on multiple factors, including the workload of bandwidth and storage, the CPU usage rate, and the protocol used. I hope this link can be helpful to you.

https://kb-cn.netapp.com/on-prem/ontap/Perf/Perf-KBs/ONTAP_9_Performance_Resolution_Guide 

Best regards