Streaming issues with my Roku when connected to my gigabit switch.

[killeraxemannic]

Ok so really my question is are all switches created equal?

I have a TP-LINK TL-SG1008D Gigabit switch in my living room. It has a 50ft cat 6 cable running to it from my Asus RT-N65R router in my computer room. When I have my Roku 3 connected to the switch I seem to get longer loading times and more buffering compared to when I connect it directly to my router. I also have a home theater PC and seem to get about 5 mb/s slower speeds when I run an internet speed test compared to my desktop which is plugged directly into the router.

Are there faster and slower switches out there or do I have another issue?

If so what switch is recommend to get good performance from the devices connected to it?

 

[azazel1024]

Ok so really my question is are all switches created equal?

I have a TP-LINK TL-SG1008D Gigabit switch in my living room. It has a 50ft cat 6 cable running to it from my Asus RT-N65R router in my computer room. When I have my Roku 3 connected to the switch I seem to get longer loading times and more buffering compared to when I connect it directly to my router. I also have a home theater PC and seem to get about 5 mb/s slower speeds when I run an internet speed test compared to my desktop which is plugged directly into the router.

Are there faster and slower switches out there or do I have another issue?

If so what switch is recommend to get good performance from the devices connected to it?

I am thinking a different issue. I don't have super extensive experience, but the experience I do have with consumer switches is, either they work or they don't, especially on a dumb switch level. I haven't ever actually experience one slowing traffic or causing appreciable delays.

Every switch you have between you and your end route WILL cause delays, but we are talking on the order of at most 1ms per switch that your traffic has to pass through. Not something you'd ever notice unless you want to stack 100 switches. That would be very, very bizzare and the first time I've seen any kind of delay being caused. The only things I can possibly think would be some kind of loop back or traffic collision going on, but the former would be more likely to lock your switch and the later is pretty darned unlikely to be a switch issue, but I guess could be.

I've also never noticed or experienced slow down with switches. I'd suspect either a difference in run to run variance on a speed test, or a difference in NICs between your HTPC and your desktop causing it (or operating systems).

 

[L&LD]

Customers have insisted I use the cheapest switches possible (google can be dangerous) because the spec's are the same.

However, for my systems and the customers that can pay a little more, I use Netgear switches exclusively.

http://www.memoryexpress.com/Catego...ock&Sort=PriceDesc&PageSize=120&ViewMode=Icon


No problems yet (except a couple of DOA's) in over a decade.

 

[azazel1024]

Correction on that latency, factor of ten lower, about 100us latency per switch for store and forward switch on fast ethernet. Gigabit connection would be lower latency (as it takes less time for the full packet to get "sucked up" by the switch before it is transmitted on to the proper recipient).

 

[killeraxemannic]

I am thinking a different issue. I don't have super extensive experience, but the experience I do have with consumer switches is, either they work or they don't, especially on a dumb switch level. I haven't ever actually experience one slowing traffic or causing appreciable delays.

Every switch you have between you and your end route WILL cause delays, but we are talking on the order of at most 1ms per switch that your traffic has to pass through. Not something you'd ever notice unless you want to stack 100 switches. That would be very, very bizzare and the first time I've seen any kind of delay being caused. The only things I can possibly think would be some kind of loop back or traffic collision going on, but the former would be more likely to lock your switch and the later is pretty darned unlikely to be a switch issue, but I guess could be.

I've also never noticed or experienced slow down with switches. I'd suspect either a difference in run to run variance on a speed test, or a difference in NICs between your HTPC and your desktop causing it (or operating systems).

A loop back would just be a looped cable plugged in to 2 ports on the switch correct? I know that's definitely not the issue.

I know the HTPC will get the same speeds as my desktop connected directly to the router because it was my desktop for a long time. It was getting the same speeds as my new desktop on a speed test when it was connected directly to the router.

How would I test for the traffic collision?

Customers have insisted I use the cheapest switches possible (google can be dangerous) because the spec's are the same.

However, for my systems and the customers that can pay a little more, I use Netgear switches exclusively.

http://www.memoryexpress.com/Catego...ock&Sort=PriceDesc&PageSize=120&ViewMode=Icon


No problems yet (except a couple of DOA's) in over a decade.

I did buy that switch because it was the cheapest one I could find with good reviews. I didn't think it would cause any problems because I was just going to be using it for the Roku, my HTPC, and my PS4 in the living room. I figured it was really only going to have one active connection at a time so it wouldn't be that big of a deal if it wasn't the highest quality.

 

[azazel1024]

Try wireshark for collisions. Its a great tool to see all kinds of stuff going on with your network traffic. You can run it on both your desktop and the HTPC and compare.

For speeds now though, can you plug the router directly in? I ask because if it used to be used as your desktop, I assume it isn't exactly the same configuration as it was previously (OS/software/hardware wise).

For a loop back, yes, same cable plugged in to two ports on the switch. The same can be accomplished by plugging two switches together, with more than one port, that don't have their ports setup for LAG or LACP.

 

[killeraxemannic]

Try wireshark for collisions. Its a great tool to see all kinds of stuff going on with your network traffic. You can run it on both your desktop and the HTPC and compare.

For speeds now though, can you plug the router directly in? I ask because if it used to be used as your desktop, I assume it isn't exactly the same configuration as it was previously (OS/software/hardware wise).

For a loop back, yes, same cable plugged in to two ports on the switch. The same can be accomplished by plugging two switches together, with more than one port, that don't have their ports setup for LAG or LACP.

I didn't change the desktop any when I moved it out. I just transferred my stuff off of it onto my new computer. There is nothing different OS wise or anything from when it was in my computer room.