That is the number of copy operations it will run at the same time so if you have bandwidth, disk performance and CPU performance to spare, it will make use of them concurrently. I think my hard drive is limited to 1200 Mbps read/write, though, so 558 Mbps is about half that, so it seems to make sense that that's all it should be able to get by reading and writing at the same time. I don't think robocopy goes through the network if it's local PC.
Bvckup copying identicle files code#
My bottleneck appears to be getting from the networking code back out to the file system (on either PC).įor test, yeah, I used robocopy from PC 1 to itself. I think for iperf it sends all the data through the networking code, the network is just super-fast in the case of PC 1 to PC 1, as evidenced by test A.1. These were run on either PC 1 or PC 2 as the server.
These were all run on PC 1, making it the client.įor the rows where it says destination is "Dump" I used "iperf3 -s" and for the rows where it says destination is "File" I used "iperf3 -s -F file.out". You seem to have a problem writing data to disk, not on the network.įor test, yeah, it is all using iperf.įor the rows where it says source is "Random" I used "iperf3 -c " and for the rows that say source is "File" I used "iperf3 -c -F a512MB.file". If that is the case, I would really look at disk performance. It seems like maybe the issue with all of them is when you are actually writing out files (rather than copying and then dumping the result at the destination. Or something you are doing is wrong, or I am just completely off on what your charts are trying to show. Right? If so, that matches your iperf test to a reasonable degree. under B, you used robocopy from PC1 to itself? And it did 558 Mbit per second? Not sure how that is supposed to work, if robocopy really goes through the whole network transaction or just realizes it is on the same PC and does a local file copy.īut from PC1 to PC2 using robocopy was 32 Megabit per second. Under C, it seems you tested copying data from PC1 to itself and a 512MB file was copied instantly, and from PC1 to PC2 (maybe?) used 800 Mbit per second, which seems fine.
If you tried to use iperf to test writing data to disk, well, I've never done that, not sure if it even can, I kinda don't think so.Īnyway, I am not super clear on what your test data is showing. Is that what you did? If so, start with looking at the hard drive, that's terrible. well that's a problem but I am not sure if I am reading your table correctly. In any case, if you can't have it write data from PC 1 to its own hard drive faster than 32 megabit per second.
Copying thousands of small individual files will be much, much slower than a single large file. The most important factor is efficiency of connection setup and the copy tool. I dunno, 10 years, should be able to move data at gigabit speed, assuming ideal conditions. Unless you believe you have particularly slow computers, you can probably throw out those numbers, anything within the last.
Bvckup copying identicle files generator#
When you say source Random, you mean you used a random value generator as a data source for the test? And the destination was the same computer, so that is why you say the result is basically the speed of the random generator (the CPU)? It sounds like tests under section A were all done with iperf? I'm having trouble reading your tables a bit.
0 kommentar(er)
