Filer statistics

Commands for displaying filer statistics

You use the sysstat command to display current filer statistics and the uptime command to display filer statistics that have accumulated since the last system boot.

About the sysstat command

The sysstat command displays statistics about the following:

CPU utilization
File server protocol operations
Network traffic
Read and write operations on disks and tapes
Buffer cache age

The sysstat output is particularly useful for studying the performance of your filer.

Depending on the applications that access the filer, the computers on the network, and the network configuration, the following suggested actions might improve your filer's performance:

If CPU utilization is high, you might need to add another filer or upgrade to a more powerful one.
If the network traffic shown in the Net Kb/s column is at or near the capacity of the network interfaces (1,000 Kb/s for 10Base-T Ethernet adapters, 10,000 Kb/s for FDDI or 100Base-T Ethernet adapters, and 15,000 Kb/s for ATM OC-3 adapters), you might need to add another network adapter card to the system.
Note
For cluster systems, the sysstat command reflects the filer that is executing the commands, not the partner.

Example

The following is an example of the default output format for the sysstat command with no options.

sysstat

CPU   NFS  CIFS  HTTP   Net kB/s  Disk kB/s    Tape kB/s   Cache

                        in  out   read write   read write  age

Display interval

By default, the filer displays statistics every 15 seconds; you can specify the interval, in seconds, as an argument to the sysstat command.

Extended utilization statistics

The sysstat -u command displays extended utilization statistics instead of the default display. This display includes the total number of operations per second for NFS, CIFS, and HTTP, cache age, cache hit percentage, the percentage of time the filer spends in a consistency point, the causes of the consistency points, and the disk utilization percentage of the busiest disk.

Note
Disk utilization is an access-based value and not the percentage of the disk used. High percentages for disk utilization might indicate a disk bottleneck. A value of 60 percent can indicate a disk bottleneck, whereas a value of 80 percent is a definite sign of one.

Sysstat statistics in extended format

The sysstat -x command displays all system activity statistics in an extended output format. This produces output that is longer than 80 columns and is intended to be displayed on a wide viewer or to be captured to a file for later analysis. For the information displayed in each column, see the man page for the sysstat command.

Filer statistics summary

The sysstat -s command displays a summary of the gathered statistics after the sysstat command terminates. The summary displays maximum and minimum values, and an average of values.

Note
Except for the average percentage of cache hits, the summary of averages for percentage values is an average of percentages, not a true mean of the data.

Example

Entering the following command displays extended utilization statistics followed by a summary:

sysstat -u -s

Command to terminate the sysstat command

You can terminate the sysstat command by either typing Ctrl-C or by setting the interval count using the sysstat -c command.

Example

The following command displays 20 samples of extended utilization statistics:

sysstat -u -c 20

About the uptime command

The uptime command prints the current time, the length of time the system has been up, and the total number of NFS, CIFS, HTTP, and DAFS operations the system has performed since it was last booted.

Example

A typical example of the uptime display is as follows:

uptime

8:54am up 2 days 22:23, 3122520 NFS ops, 130 CIFS ops, 62 HTTP ops, 
0 DAFS ops

Note
For cluster systems, the uptime command reflects the filer that is executing the commands, not the partner.