Operating System - HP-UX
1752587 Members
4250 Online
108788 Solutions
New Discussion

how much bandwidth has been used by NFS?

 
SOLVED
Go to solution
Hanry Zhou
Super Advisor

how much bandwidth has been used by NFS?

We have 1GbE pipe, and shared by different purpose. One of them is NFS connection. How do I determine how much bandwidth has been used by NFS?

 

Thanks!

none
3 REPLIES 3
Dave Olker
HPE Pro

Re: how much bandwidth has been used by NFS?

What types of NFS clients and servers are you using?  HP?  NetApp?  Linux?  What OS are they running?

 

Dave

I work for HPE

[Any personal opinions expressed are mine, and not official statements on behalf of Hewlett Packard Enterprise]
Accept or Kudo
Hanry Zhou
Super Advisor

Re: how much bandwidth has been used by NFS?

We are using NetApp as the NFS server, and 4 HPUX 11iv2 as NFS clients.

none
Dave Olker
HPE Pro
Solution

Re: how much bandwidth has been used by NFS?

There's no really good way to pinpoint exact amounts of NFS traffic that I know of.  You can estimate it by using nfsstat on both client and server.  

 

From the client side I'd begin by zeroing out the counters via nfsstat -z and then pausing for a few seconds and issuing nfsstat -c.  That will give you a count of the number and type of NFS requests that occurred during that interval.  You'd have to know what size of read/write requests are occurring to estimate how much actual NFS packet data is going over the wire.

 

On the NetApp side, things are a little easier because they actually keep track of the number and size of requests coming into the filer.  For example, on my filer, I issued an nfsstat -z to zero the counters and then issued an nfsstat -s to dump out the counters:

 

 

Server nfs V3: (5034206 calls)
null getattr setattr lookup access readlink read
41 0% 12544 0% 11498 0% 40583 1% 51965 1% 0 0% 4394014 87%
write create mkdir symlink mknod remove rmdir
477398 9% 1309 0% 358 0% 0 0% 0 0% 555 0% 95 0%
rename link readdir readdir+ fsstat fsinfo pathconf
901 0% 0 0% 0 0% 41814 1% 1102 0% 25 0% 4 0%
commit
0 0%

Read request stats (version 3)
0-511 512-1023 1K-2047 2K-4095 4K-8191 8K-16383 16K-32767 32K-65535 64K-131071 > 131071
87 5803 17459 10532 148425 64829 46049 201967 3898863 0
Write request stats (version 3)
0-511 512-1023 1K-2047 2K-4095 4K-8191 8K-16383 16K-32767 32K-65535 64K-131071 > 131071
265322 11187 6182 4871 115281 39925 11604 2586 20440 0

 

 

The format in this page probably looks horrible but the real output on the filer shows me how many NFS requests of each type were received during the interval and it breaks down the size of read/write requests into buckets of 0-512bytes, 512-1K, 1K-2K, 2K-4K, etc.  Knowing how many packets and the size of the payload can give you a lot of hints as to how much NFS traffic is coming into the filer on that link.

 

Good luck,

 

Dave

 

I work for HPE

[Any personal opinions expressed are mine, and not official statements on behalf of Hewlett Packard Enterprise]
Accept or Kudo