vBulletin Search Engine Optimization
| |||||||
| Register | FAQ | Members List | Calendar | Search | Today's Posts | Mark Forums Read |
| |||
| Tommy Fallsen wrote: > Hi > Anyone had performance problems with this nic? > > > Tommy Fallsen > > Most gigabit performance problems are because the host driving it is insufficient or the test is flawed. What platform are you running on and what performance are you getting? What tests are you doing to determine that performance? Elias |
| |||
| "Elias" <nospam@charter.net> wrote in message news:vmnc6cb0bisl73@corp.supernews.com... > Tommy Fallsen wrote: > > Hi > > Anyone had performance problems with this nic? > > > > > > Tommy Fallsen > > > > > > Most gigabit performance problems are because the host driving it is > insufficient or the test is flawed. > > What platform are you running on and what performance are you getting? > What tests are you doing to determine that performance? > No test done yet. Not even sure of a good way this is done. It just that this is the only 280R clearcase server with cassini and user complain about it being slow and netstat,vmstat,sar,nfsstat,iostat indicates the opposite. Tommy Fallsen |
| |||
| Tommy Fallsen wrote: > "Elias" <nospam@charter.net> wrote in message > news:vmnc6cb0bisl73@corp.supernews.com... > >>Tommy Fallsen wrote: >> >>>Hi >>>Anyone had performance problems with this nic? >>> >>> >>>Tommy Fallsen >>> >>> >> >>Most gigabit performance problems are because the host driving it is >>insufficient or the test is flawed. >> >>What platform are you running on and what performance are you getting? >> What tests are you doing to determine that performance? >> > > No test done yet. Not even sure of a good way this is done. It just that > this is the only 280R clearcase server with cassini and user complain about > it being slow and netstat,vmstat,sar,nfsstat,iostat indicates the opposite. > > Tommy Fallsen ttcp with multiple streams is the best way to test the throughput of your network connection. This is covered in previous postings. There are CE patches that you can load. Let's assume Solaris 8. If you don't have at least 108528-17 then there are CE patches that you are missing in the kernel (108528-23 is the latest). Also 111883-18 and 112119-04. If you can, make sure the card is in the 66MHz PCI slot. Are any of your other system resources being maxed out? CPU, memory, disk? Is everything else about the other servers identical except for the NIC? Elias |
| |||
| Tommy Fallsen <tommyf@kongsberg.com> wrote: > No test done yet. Not even sure of a good way this is done. It just > that this is the only 280R clearcase server with cassini and user > complain about it being slow and netstat,vmstat,sar,nfsstat,iostat > indicates the opposite. When you do run some network perf tests, make sure to include some latency measures like the netperf TCP_RR test. Some things can be rather more latency sensitive than bandwidth sensitive. rick jones -- denial, anger, bargaining, depression, acceptance, rebirth... where do you want to be today? these opinions are mine, all mine; HP might not want them anyway... feel free to post, OR email to raj in cup.hp.com but NOT BOTH... |
| |||
| Rick Jones <foo@bar.baz.invalid> writes: > Tommy Fallsen <tommyf@kongsberg.com> wrote: > > No test done yet. Not even sure of a good way this is done. It just > > that this is the only 280R clearcase server with cassini and user > > complain about it being slow and netstat,vmstat,sar,nfsstat,iostat > > indicates the opposite. > > When you do run some network perf tests, make sure to include some > latency measures like the netperf TCP_RR test. Some things can be > rather more latency sensitive than bandwidth sensitive. A good example is Visual Source Safe. It cannot be made to perform well over a transatlatic link even you have god's own bandwidth. -- Chris Morgan "Post posting of policy changes by the boss will result in real rule revisions that are irreversible" - anonymous correspondent |
| |||
| > >>Most gigabit performance problems are because the host driving it is > >>insufficient or the test is flawed. > >> > >>What platform are you running on and what performance are you getting? > >> What tests are you doing to determine that performance? > >> > > > > No test done yet. Not even sure of a good way this is done. It just that > > this is the only 280R clearcase server with cassini and user complain about > > it being slow and netstat,vmstat,sar,nfsstat,iostat indicates the opposite. > > > > Tommy Fallsen > > > ttcp with multiple streams is the best way to test the throughput of > your network connection. This is covered in previous postings. > > There are CE patches that you can load. Let's assume Solaris 8. > > If you don't have at least 108528-17 then there are CE patches that > you are missing in the kernel (108528-23 is the latest). > Also 111883-18 and 112119-04. It's running 108528-15, but will be upgraded to 108528-23 tomorrow because of Securepath upgrade. I installed 111883-18 last week, did'nt help, and 112119-04 failed. But the cassini will get replaced by a ge interface. Got one spare and have permission to take the server down anyway. > > If you can, make sure the card is in the 66MHz PCI slot. All network interfaces is installed in this slot by default. > > Are any of your other system resources being maxed out? CPU, memory, > disk? Is everything else about the other servers identical except for > the NIC? They are all the same, but SF280R with cassini nic is the fastest one with 1050Mhz cpu's. Tommy Fallsen |
| ||||
| Chris Morgan <cm@mihalis.net> writes: >Rick Jones <foo@bar.baz.invalid> writes: >> Tommy Fallsen <tommyf@kongsberg.com> wrote: >> > No test done yet. Not even sure of a good way this is done. It just >> > that this is the only 280R clearcase server with cassini and user >> > complain about it being slow and netstat,vmstat,sar,nfsstat,iostat >> > indicates the opposite. >> >> When you do run some network perf tests, make sure to include some >> latency measures like the netperf TCP_RR test. Some things can be >> rather more latency sensitive than bandwidth sensitive. >A good example is Visual Source Safe. It cannot be made to perform >well over a transatlatic link even you have god's own bandwidth. Presumably, God could change "C" :-) It's an eternal fight between our networking staff and the business apps developer; the latter don't seem to grasp that it's the number of (synchronous) transactions that is all-important, and not the number of bytes which is largely irrelevant. Casper -- Expressed in this posting are my opinions. They are in no way related to opinions held by my employer, Sun Microsystems. Statements on Sun products included here are not gospel and may be fiction rather than truth. |