March 29, 20179 yr I'm using a single NIC on my unraid server but experiencing strange performance, when doing a single read from a share I'm able to max out the speed of a drive, but if I start a write operation, to a different drive not in the array (or vice versa), the performance takes a huge nosedive, then transfer that is already in progress slows right down and the send to the unraid share crawls. For the sake of argument, reading from //tower/domains (which is stored on the cache drive) sits between 80 and 90MB/s, then starting a send operation to //tower/images (which is on a non-parity protected array) kills the performance of the current transfer, and the send operation crawls along. If either transfer is cancelled, the other one speeds up. Could it just that that I'm running out of bandwidth on the Intel embedded raid ii controller? You can see in the attachment where the send operation was started at 15:42:00 then once the receive operation finished at 15:42:30 it sped up considerably. Iteration 1 was whilst duplex was going on. Running a 400MB file write on Z: 5 times... Iteration 1: 13.98 MB/sec Iteration 2: 111.69 MB/sec Iteration 3: 117.65 MB/sec Iteration 4: 103.73 MB/sec Iteration 5: 111.27 MB/sec ----------------------------- Average (W): 91.66 MB/sec ----------------------------- Running a 400MB file read on Z: 5 times... Iteration 1: 105.32 MB/sec Iteration 2: 100.21 MB/sec Iteration 3: 106.55 MB/sec Iteration 4: 102.79 MB/sec Iteration 5: 103.47 MB/sec ----------------------------- Average (R): 103.67 MB/sec ----------------------------- Edited March 29, 20179 yr by Spies
Archived
This topic is now archived and is closed to further replies.