Performance Monitoring

[Guest]

Hi All,

I am playing with the idea of upgrading our production web servers to
gigabit Ethernet as a number of them are creating large amounts of network
traffic during peek periods. Firstly I wanted to see if there were any
bottlenecks on the servers by testing their network interfaces. I read on a
number of websites that monitoring the 'Output Queue Length' counter in
Performance Monitor is a good place to start. So I decided to monitor this
counter on a number of servers at a time when there is very little traffic on
the sites (around 10 users). I instantly noticed that the counter readings
were around the 42 million mark for those systems that have a direct
connection to the SQL servers (using a crossover cable) and around the 300
mark on a number of other servers which connect to our backbone network via a
number of segmented switches. According to a number of websites, the reading
for the above counter should remain around the 0 mark.

I would normally put this down to a faulty network card or driver, but as
the problem is affecting a number of machines that don't all sit on the same
network I am very puzzled.

I would be most grateful if anyone has any ideas, thanks in advance.

Windows 2000 Servers (SP4)
SQL 2000 (SP3)
Dell PowerEdge 2650
Broadcom Network Extreme gigabit Ethernet adapters
3COM and Cisco 10/100 switches

 

[dave]

I am playing with the idea of upgrading our production web servers to
gigabit Ethernet as a number of them are creating large amounts of network
traffic during peek periods. Firstly I wanted to see if there were any
bottlenecks on the servers by testing their network interfaces. I read on a
number of websites that monitoring the 'Output Queue Length' counter in
Performance Monitor is a good place to start. So I decided to monitor this
counter on a number of servers at a time when there is very little traffic on
the sites (around 10 users). I instantly noticed that the counter readings
were around the 42 million mark for those systems that have a direct
connection to the SQL servers (using a crossover cable) and around the 300
mark on a number of other servers which connect to our backbone network via a
number of segmented switches. According to a number of websites, the reading
for the above counter should remain around the 0 mark.
I would normally put this down to a faulty network card or driver, but as
the problem is affecting a number of machines that don't all sit on the same
network I am very puzzled.
I would be most grateful if anyone has any ideas, thanks in advance.

Hello,
I'm not quite sure about what kind of monitoring tool you are looking for.
Maybe you should try to do it with 30 trial version of some tool. A lot
vendors make somethings like this accessible with their tools, i.e: adRem
with their netcrunch (network monitoring tool). You can dowload it from
theirs website or try to find it over the intrnet.

dave