Issue reared it's head when my cust doing file transfer with win2k3 server to XP pc pack 2 client using drag and drop . The two consoles are directly connect to same 3560 switch.Server to copper sfp port running 1000 full client to fast e @ 100 full. routing is rip .is intervlan but other test work ok ie 100 to 100 fast e intervlan same procedure also gig to gig . did trace and shows higher layer response time @ 132msec compared to fast transfer @ 2ms higher layer response times. Same type of issue for another cust with sfp mod's. worked fine with GBIC's. Cisco says thier test shoeww no issue although stopped testing with XP pack 1 client. bottom line is finding out why response times so high for upper layer protocol NBSS.
A big reason for slower than expected throughput could be the size of TCP receive window size. I believe that Solaris defaults to a window size of 8K bytes. As they increase the window size, they should see a proportional increase in throughput
Learn, share, save
Discover and save your favorite ideas. Come back to expert answers, step-by-step guides, recent topics, and more.
