How to identify a network card bottleneck

[Edinfor]

Hi All,

Is possible to make some test to identify if a bottleneck is in a network card?

Thanks,

Nascimento

[Hakan Aribas]

Hi,

You can use the following command:
netstat -p tcp

Some general tips:
* Make sure your network card is running full duplex. Some auto-negotiation protocols
with the network hardware may inadvertently set your card into half-duplex mode. Avoid autonegotiation when possible.

* Use of the automounter can cause unproductive flushes of data from the buffer cache. If possible, use AutoFS instead of Automount.

* If you need NFS, use NFS V3. Clients need to be talking V3 as well as the server.

[Jeff Schussele]

Hi,

The best tool to use is MeasureWare & it's main component PerfView.
But, these are "pay-per-view" products.
But, seriously, they're well worth it.

My 2 cents,
Jeff

[rick jones]

While netstat -p tcp will give you good statistics about what is happening with TCP on the system, in and of itself it cannot tell you that you have a bottleneck at your NIC. At most it tells you if packets are being lost, in which case there may be a bottleneck somewhere, or no packets are being lost, in which case any bottleneck is somewhere besides the "network" itself.

Glance and Measureware can track a number of link-level statistics which can indicate a NIC saturation - for example, if the outbound transmit queue of the NIC stays "high" for a prolonged period of time (for subjective definitions of high and prolonged) they will flag the NIC as a possible bottlenect.

If there are inbound or outbound discards on the NIC, that too can be flagged (and may appear as retransmissions in TCP).

You can look at the lanadmin statistics yourself if you prefer:

lanadmin -g mibstats

Related to a NIC bottleneck, glance and measureware can tell you if a specific CPU is "pegged" - ie completely consumed - with interrupt time. You can correlate that with the information from the intctl command to see which NIC is interrupting which CPU.

[Steven E. Protter]

http://www.hpux.ws/system.perf.sh

The Non pay per view version. The network management toolset here is pretty good. It will identify issues involving the network and other possible bottlenecks.

Glance is better, but this is free.

SEP

[morganelan]

Important early step in troubleshooting any Network performance issues:
#How many network hops (i.e. bridges, hubs, routers, switches, etc.) do network packets traverse between the client and the server systems?
#What is the speed of each link separating these systems?
#Does your network equipment use auto-negotiation to set speed and duplex settings?
#Are your network interfaces configured for half-duplex or full-duplex mode?
#Do your switch port settings match the speed and duplex settings of your host interfaces?
#What is the maximum transmission unit (MTU) size of the links between these systems?
#If the links are using different MTU sizes, how are the packets being translated? For example, if the NFS client resides in an FDDI ring and uses an MTU size of 4352 and the NFS server uses a 100BT interface with an MTU size of 1500, how are the 4352 byte packets from the client being fragmented into 1500 byte packets for the server?
#Do packets sent from the client to the server take the same route through the network as the packets sent from the server to the client?

You can man netstat to find what you want to know about your network statistics.The traceroute(1M) tool provides a simple means of determining the path through the network taken by packets sent from one system to another.Another tool shipping with HP-UX that can simplify the process of collecting network topology information is ping(1M).The lanadmin command allows a system administrator to display many useful statistics kept by the LAN driver subsystem, regardless of the interface type.

You can try using tools such as:
ttcp: http://ftp.arl.mil/ftp/pub/ttcp
netperf:http://netperf.org