Performance

For performance testing, I have been using the ECS2512 in my office so I could do testing between test benches and also to eliminate any wire issues from our Cat5e wiring when doing tests. Before getting into those numbers however with the switch in my office I did get some noise testing as well. The ECS2512 is designed and intended for use rack-mounted but with it having active cooling in my otherwise quiet office I would check out the numbers. It came in at 45.1 decibels when testing in the A-weighted test. Then in the c-weighted test, the ECS2512 came in at 58.3 decibels. Without testing the ECS2512FP I can still say that this is quieter with it having just one fan rather than two, but this is not a switch designed to hide in the background in a quiet room.

One big focus with a switch, especially if you plan on pushing its limits is to make sure the overall switching capacity is enough to handle things. For the ECS2512 that would mean that it needs to be able to handle the eight 2.5Gbps ports in duplex which would be 5Gbps per port or 40Gbps in total. Then we also have to include the four SFP+ ports as well which are 10Gbps each in duplex for 80Gbps or 120Gbps in total which is exactly what the ECS2512 has listed in its specifications.

For testing, I had a good idea of what to expect given how similar the ECS2512 is to the ECS2512FP. To test the network performance I ran Passmark Performance Test 10 on two PCs which both have 2.5G and 10G copper network cards in them. I ran a variety of tests and included the min and max results as well. I tested 10G to 10G when running both on 10G SFP+ ports using THIS SFP+ to RJ45 Copper Module from ADOP on the cheap previously. I tested again then with one PC hooked to the Intel I225-V NIC for 2.5G. I did this same test again with one hooked to 10G on the ECS2512 and one to 2.5G to check performance between the main switch and the SFP+ ports and again with both on 2.5G ports as well. There weren’t any big surprises and a lot of the tests are almost the same result except for the 10G to 10G test. There was one 10G to 10G nic test when going from SFP+ to 2.5G that was just slightly lower than the rest but not enough to be a concern at all.

Average Speed

Minimum Speed

Maximum Speed

Both on 10G SFP+ Ports

Intel I225-V to  Marvell 10G

2320 Mbits/Sec

1631 Mbits/Sec

2437 Mbits/Sec

Marvell FastLinQ Edge 10G to  Marvell 10G

9470 Mbits/Sec

8968 Mbits/Sec

9531 Mbits/Sec

One on 10G SFP+ and one on 2.5G Switch Port

Intel I225-V to  Marvell 10G

2358

1820

2375

Marvell FastLinQ Edge 10G to  Marvell 10G

2333

1735

2373

Both on 2.5G Switch Port

Intel I225-V to  Marvell 10G

2355

1844

2375

Marvell FastLinQ Edge 10G to  Marvell 10G

2358

1822

2375

 

Log in to comment

We have 1222 guests and one member online

supportus