I've recently set up link aggregation on my home network but I'm not getting an increase in throughput. I have a server running linux mint with a dual port intel pro 1000 pt nic, which is connected to switch A (TP-Link TL-SG3216), switch A is connected to Switch B (HP PS1810-8G) and on switch B I have PC1 and PC2. To test it I used both PC1 and PC2 to copy files from the server but I was limited to a single link speed. I'm using LACP and as far as I can tell everything looks like it's working properly.
I haven't tested what happens if the server and the two PCs are on the same switch but I was expecting that if the switches were linked using two ports with link aggregation then I would be able to get a higher throughput from the server.
Right off the top of my head... depends on what medium you're copying those files to... right? Even at single gig speeds your bottleneck is going to be the write speed of your disk(s)
Both computers are using SSDs and i'm copying off different disks on the server which can read around 200MB/s. When I have both PCs copying the server is uploading around 120MB/s and when I stop one it stays the same. This test doesn't really represent any real world usage it's more just to see if everything is working properly.
I guess my question is; if I should get more than a single link speed from the server when more than one computer is accessing it then is that still the case when the server is connected to one switch and the computers are connected to another? Or in other words does link aggregation between switches allow link aggregation between the devices on all the switches or just the switch that the devices are connected to?
The switch needs to be LAGG capable, and also you need to set it up on the switch as well as the server and pc. Also if you are running between multiple switches they all need the same cabling as the devices it is going to/from. I.E. If you are running a dual NIC card, Then you need 2 cables all the way from point of origin to termination.
Yeah it's all set up properly. I have a dual nic on the server with two cables to the first switch, and two cables from that switch to the second switch. link aggregation is set up and working on the switches and the interface on the server.
My expectation is if this is the set up: server > switch A > switch B > PC 1 and PC 2, that if i transfer files between the server and both PC 1 and PC 2 at the same time, I should get more than 1gbps speed, but when I try this I only get 1gbps. So I don't know if this is how it is supposed to work or if something isn't configured properly.
Yes, the server has two gigabit connections to the first switch which has two connections to the second switch. When I transfer files to both computers on the second switch I only get 1gbps total bandwidth. I get the same bandwidth if only one computer is transferring. What I want to know is should I get more than 1gpbs total bandwidth with this configuration or does having the server and the client PCs seperated by a second switch (even though that switch is connected to the frist with link aggregation) cause the total bandwidth to only be as fast as a single gigabit link?
Another observation I've had while testing this is that when I transfer files, even to a single pc, both aggregated links show activity.
With link aggregation each computer's maximum bandwidth is still limited to a single link speed, so around 120MB/s for gigabit. Even if both computer have dual nics. The benefit is when you have multiple computers accessing a server, say you have two computers then both will get 120MB/s, if you have four then they will get 60MB/s etc.
I think there is a way to double your bandwidth using a link aggregation mode in linux in which the network traffic is striped similar to RAID 0 but it's unreliable and suffers from collisions and packets arriving out of order.