February 5, 2003 at 3:42 pm
I have a simple select * from a table with 300,000 rows. In query analyzer on the server this takes 1.1 minutes. On client machines it takes about the same. As of thursday, however I have 1 terminal services (in application server mode) machine, that is also a domain controller. The same query takes over 9 minutes. All machines are using tcp/ip 1433 and named pipes for protocals. The terminal server box is using less than 10% of its resources (p4 2GHZ 512mb Ram, 120 GB disk)
All machines are windows 2000 sp3 (server and workstation) The sql server is sp3 and so is the client. I also see some dsn's that I cannot delete as I am told they are invalid.
Any ideas?
February 5, 2003 at 4:17 pm
Check the NIC input/output packets. May be high and thus packets are blocking or being lost so they have to retransmit. ALso, try pinging from that server and see what the average response time is, then do a tracert to get statis from the network movement and compare to another near that machine on as close to the same network segment as possible.
February 5, 2003 at 5:32 pm
I tried each even looked at MTU's. When I look at the switch there is no activity between the server and the TS box. However when I run the same query on a different box my switch shows lots of activity between the boxes. I tried switching ports, cables, all nics are set at 100 full duplex as are the switch ports. Its like the data just aren't making it to the nic.
February 6, 2003 at 3:39 am
I would suggest to confirm what is happening on the box, install Netmon and monitor all incoming and outgoing packets then run the query from that box. Once the data appears take a look at the packets to see if anything unusual is appearing. I am sure there are other possiblities but I cannot think of them so if someone else will join in with options please do so.
Viewing 4 posts - 1 through 3 (of 3 total)
You must be logged in to reply to this topic. Login to reply