I’ve set up my NAS with Link Aggregation on a compatible router but I’m not getting the throughput expected, getting 113MB/s max.
Running a Synology DS1515+ (with 5 disks in SHR and 6 gb ram) with 2x1Gb patches to NetGear GS108Tv2
For the switch I followed the instructions to set up the LAG Configuration:
Lag ID 1, Link Trap=Disable, Admin Mode=Enable, STP Mode=Disable, LAG Type=LACP, Active Ports=g6 LAG state=up
Set LAG Membership to ports 6 and 7.
I set up the NAS side, first setting up the two nic’s with static IPs on the same subnet, then “bond” them using IEEE 802.3ad Dynamic Link Aggregation, and it shows all green, but “Network Status” shows only 1000Mbps, Full Duplex, not 2000Mbps as it should.
Lan1, 1000Mbps, Fulll Duplex, MTU 15000
Lan2, 1000Mbps, Fulll Duplex, MTU 15000
Network Status shows 1000Mbps, Full Duplex, MTU 1500
When I test by copying a file to one PC and then the same or different file to a different PC, the 113MB/s bandwidth gets split between the two.
This does indicate a 2000 Mbps aggregate link speed.
I also use the GS108Tv2 managed switch. Is 113 MB/sec the observed peak or sustained speed? Does the PC have 1 or 2 network ports? If only 1 port, you have reached the limit of what the single segment on the PC side of the switch can handle as it is only 1000 Mbps. If 2 ports, you need to verify its LACP configuration and LAG in the GS108Tv2. It will be LAG ID 2.
The best method of confirming performance is with 2 or more devices (single ethernet port) connected to the switch. If both can pull 100 MB/sec, you are indeed getting the 2 Gbps LACP speed.
Hello thanks for the quick response.
I’m testing with two different PC’s. To test I start copying a file from the NAS to one PC, then start copying a file from the NAS to the 2nd PC. PC 1 starts at 113MB/s but then drops off as soon as I start the 2nd transfer.
I’m sure it has something to do with the NAS setup as it only says 1000 Mbps under the network status.
Ok, so I figured out why I wasn’t getting the 2000 Mbps showing up, I had mixed up the cables, plugged in the correct one and set it up again, boom, 2000 Mbps showing in Synology. And ports 7 and 8 showing up as “active” in LAG configuration.
But then when I test with a file transfer… same as before, PC#1 starts at 113MBs and then drops top 50-60 as soon as I start the dl on PC #2. If I stop the copy on PC#1, PC#2 goes up to 113MBs.
No but now you have me questioning my sustained throughput. Since the Synology update last week, I’ve seen my single-segment copy speed (read-process-write back) dropping into the 60’s (60R / 60W). I am smelling something disturbing
I’ve switched back and tested copy speeds with just one network connection and then back again, no difference.
Link Aggregation seems to be up and running, but it’s just not working like it should, not sure if the issue is with the switch or the NAS.
Independent upload, single segment = 98+ MiB/sec concurrent with independent download, same segment, ~100 MiB/sec (fluctates a bit)
Everything is fine until I go to computer 2 and start a download of the same source file.
This is a DSM issue or we are both 100% wrong on how to do link aggregation with the GS108Tv2
Upload:
[chuck@lizum ~/nas.104]$ date; dd if=/dev/zero of=/synb/zeros.tmp bs=1M count=50000
Thu Dec 29 14:10:00 EST 2016
^[50000+0 records in
50000+0 records out
52428800000 bytes (52 GB, 49 GiB) copied, 527.721 s, 99.3 MB/s
[chuck@lizum ~/nas.105]$
Download: (ran for 3 minutes before starting second host)
[chuck@lizum Chappie (2015) [uhd].104]$ date; dd if=Chappie\ \(2015\).mkv of=/dev/null bs=1M
Thu Dec 29 14:10:01 EST 2016
49581+1 records in
49581+1 records out
51989725754 bytes (52 GB, 48 GiB) copied, 731.449 s, 71.1 MB/s
[chuck@lizum Chappie (2015) [uhd].105]$
As I write this, the other host is still reading the file. The observed lowest read-rate was 49 MiB/sec. This is entirely consistent with a host not utilizing both links of a LACP bond set.
Update. I just did a full rebuild of the LACP in both the switch and the Syno. This is definitely a Synology issue. If you pull one of LACP members and place it in another port, it comes up as ‘just another ethernet port’. This is correct behavior if it’s not in a LAG. This confirms DSM is seeing the LAG but is not aggregating (sending/receiving packets from both segments simultaneously from the same source) . You can demonstrate with 2 sources, there is a problem. “Hello, Synology?”
I rebuilt it as it was initially built. Since I hadn’t been slamming it for some time, I thought the slowdown was perhaps this computer. Given both machines are recent (last 60 days) rebuilt from scratch, now, with the LACP rebuilt too. It’s blatently obvious where the 'mail isn’t getting delivered"
When sending to the NAS from different PC’s, one started off at 110MB/s, the other started at 24 and climbed to 104MB/s, both finished off in the +100MB/s range.
Defiantly looks like an issue with the Synology side of the network.
Also, this is what I see when copying the same file to two machines simultaneously. 49 GB file, copied twice. both machines started within 2 second of each other… NAS has 8GB of ram.
old style does balance… once it’s had time to get going. It is slow to adapt though. still not stable in the reporting but that might just be their silly sampling all fowled up again
