Toss Out or Donate Old Switches

coffeecup25

I recently upgraded my ISP from cable to fiber with another company. It costs less and I am now at 1gb symmetrical. Afterward I still noticed a little slowness that seemed odd so I upgraded a switch to the newest model unmanaged switch from a 10+ year old V 1.0 smart switch.

The new switch is great. Lag is almost instantaneous. The problem I noticed is gone. Improvements over the past decade must have made the difference.

Now, what to do with the old switches. I have 5 decade+ old switches to dispose of. Some were replaced. Some were on the shelf and won't be put back into use.

Toss Out or Donate?

The old switches work well but would someone who bought from a thrift shop be disappointed with lag time that new tech apparently does not have? TP-Link and Netgear. Most are unmanaged. Port counts ranges from 5 to 24.

The old models are either end-of-life or many versions old compared to new ones. Amazon is very low priced with new models and they go on sale now and then.

Suggestions?

scottjh1

You could reset them to factory and donate them to programs that are supporting kids or senior citizens who have limited cash for hardware. Possibly the local volunteer fire company may be a good choice.

johnpoz

@coffeecup25 said in Toss Out or Donate Old Switches:

with lag time that new tech apparently does not have

And what is the lag time you saw.. I find it hard to believe a gig switch would have any lag other than a few ns.

So your saying you ping something between 2 ports and you saw what for response? And then you use the new switch and you say you saw something less? Did you go from 1 ms to 0.5 ms? Did you go from 10ms to <1 ms?

What is this lag you say your 10 year old switch was having?

Here is ping from my pc to Pi I have that goes like this

pc - switchA (sg300-28) -- pfsense - vlan -- switchA (again) - switch B (sg300-10) -- switch C (tplink tl-sg108) - pi

So not only does the traffic get routed, and through the same switchA twice, but then another switch and then a old tplink. And all of the switches are quite old.. The tplink is from 2017.. The sg300-10 is from 2014, the -28 is from 2017

From 192.168.3.10: bytes=60 seq=0001 TTL=63 ID=fede time=0.893ms
From 192.168.3.10: bytes=60 seq=0002 TTL=63 ID=feee time=0.644ms
From 192.168.3.10: bytes=60 seq=0003 TTL=63 ID=ff57 time=0.854ms

coffeecup25

@johnpoz

You know, I didn't go about scientific measurements. I noticed that I had a slightly faster response time on the home network after replacing a 24 port circa 2012 smart switch with a new unmanaged 24 port switch. It was enough for a heavy user to notice. Even my Roku seemed a little more responsive.

That switch replacement was experimental because the fiber upgrade still seemed a little laggy even though Speedtest said my internet was 2x faster. Afterward I replaced another switch and added a 5 port spare just because it only cost $13. Again, enough improvement for me to notice.

That leaves 5 overall to dispose of. I decided to donate because 1) Not all switches are used by interactive applications and 2) Not all interactive users need instant response. I didn't until I took the time to really look into it. Now I'm spoiled.

johnpoz

@coffeecup25 maybe your old switches were coming up 10 or 100mbps and not gig.

But switches do not add latency in any amount that would be noticeable.. The lag on a switch is going to be in the nanoseconds..

As far as differences between a 10 year old switch and today.. It still would only be ns.. Not anything that you could possibly notice..

coffeecup25

@johnpoz

Well, the only thing that changed was the switches and I very definitely noticed a difference. We're debating almost instantaneous vs almost instantaneous plus a half fraction of a second. I noticed it. Maybe it's a Type-A vs Type-B personality thing?

I looked into it a little before spending $80 on the new 24 port switch. Gamers, apparently, obsess on little differences they encounter. I must share some of those traits although I'm not a gamer. Switches are a big deal for them, or so the internet implies. Also, I saw references to technological improvements over time, but they were all vague.

And, no, I went from 500mb cable to 1gb fiber. Speedtest confirmed.

dennypage

@johnpoz said in Toss Out or Donate Old Switches:

I find it hard to believe a gig switch would have any lag other than a few ns.

Much, much more than a few ns.

A standard gigabit switch by definition has a minimum latency of 51.2us, and commonly a 57.6us (adding preamble and start frame delimiter).

Even a cut-through gigabit switch would still have an absolute minimum latency of 4.8us, and commonly 9.6us.

michmoor

@dennypage always chiming in with the scientific data

Patch

@dennypage said in Toss Out or Donate Old Switches:

A standard gigabit switch by definition has a minimum latency of 51.2us, and commonly a 57.6us

Thanks for providing actual figures.

But that is still several orders of magnitude below human perception which does not at all support the ops claim. Even after allowing for multiple network transmissions to fulfil a users request.

Happy the op upgraded their network and is happy with their purchase. They will probably save some power. But I still think it very unlikely anyone else would notice any difference in response time of normal activities over a LAN buy replacing a 12 year old network switch.

I suppose it maybe possible to mis-configured a programmable switch. Replacing it with a non programmable switch or resetting the programmable switch to factory defaults would correct the configuration error.

johnpoz

@dennypage said in Toss Out or Donate Old Switches:

Much, much more than a few ns.

Thanks for info ;) I mixed up nano and micro.. heheh

coffeecup25

@Patch said in Toss Out or Donate Old Switches:

But that is still several orders of magnitude below human perception which does not at all support the ops claim. Even after allowing for multiple network transmissions to fulfill a users request.

I know what I see. Especially on my Roku response time. It's an instant faster after I press the select button.

It's possible the smart switch would have performed better with a factory reset. But I never configured it in any way so, in theory as you like to think, it should have not made a difference.

I guess since you can't notice things like this, nobody else can. Annoying. My main interest was try to figure out if I should toss out or donate. The goofiness here made the whole vibe into something I won't try again.

This is much like my past post where I wanted to know about experiences with T-Mobile for WAN redundancy or as a main ISP. Some goof turned it into a long long long VLAN debate because of a configuration issue I mentioned. I tried to be polite and that seemed to be a red flag that egged him on to more long long long debate. He was good at being difficult to ignore. I know, this is the internet and that is what to expect. Common here, apparently.

scottjh1

I have a Linksys LGS318 switch. A ping on the internal network is shown below. I got it in 2014 when they first came out. It is patched with the latest version of firmware. One item to note is flow control default is set to auto, which I did change to enable. Doing so reduced the response time both internal and out to the internet. The switch internal DNS and DHCP is turned off being my Pfsense box handles those tasks.

Pinging 192.168.1.51 with 32 bytes of data:
Reply from 192.168.1.51: bytes=32 time=1ms TTL=64
Reply from 192.168.1.51: bytes=32 time=1ms TTL=64
Reply from 192.168.1.51: bytes=32 time=1ms TTL=64
Reply from 192.168.1.51: bytes=32 time=1ms TTL=64

Ping statistics for 192.168.1.51:
Packets: Sent = 4, Received = 4, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
Minimum = 1ms, Maximum = 1ms, Average = 1ms

johnpoz

@coffeecup25 said in Toss Out or Donate Old Switches:

I know what I see.

Not saying you didn't have some issue - but it sure not related to latency introduced by a switch. You shouldn't of mentioned that to be honest if you didn't want to discuss it.

Here is the thing - if it wasn't challenged, some poor slob reading this would be hey.. My ping time is X maybe its because my switch is Y years old.. Next thing you know you have users parroting switch older than X years old add huge amounts of latency..

Kind of like the nonsense you see that that to do gig you need cat 6, and that cat 5 can only do 100mbps.. I see this nonsense almost daily.

You also should be prepared with some numbers... I have been in this biz for many many years, and one thing that normally bugs anyone in the it field the most is "it's slow" -- what is slow?? Your browser loading a page now takes 30 seconds vs before 5 seconds.. Your normal file copies you to take seconds now take minutes?

If you feel its the switch - then lets see some pings across the old switch, then with the new switch as a start.

You mention roku - I have a bunch of them.. Most of them are wireless, you have a wired one - like an ultra or something, I have one its only 100mbps? Most of those only have 100mbps interfaces anyway, the old wired model was the same way, which roku(s) do you have? So switching to new gig switch wouldn't do anything.. Unless the old switch was having some problem, maybe came up 10mbps - or maybe you had some uplink from one switch to another was only 10?

Maybe your roku just need a interface rest? which would of happened when you plugged it into a different switch?

If you just wanted to ask if you should donate a switch or e-waste it.. Should of just said hey have some old dumb gig switches, I am replacing them.. What should I do with them.. Vs bringing up what makes no sense that latency in a 10 year old switch compared to today switch is some how noticeably different that can be noticed by user..

See it today, users asking if they daisy chain a switch will their internet be slower.. Like the few micro seconds that might add could some how make their internet game faster, or if they replace their cable with cat 8 will it be faster, etc. They are not doing (HFT) stock transactions, with a hardline into the stock exchange. Yeah their are some very specialized gear for that to trim latency with measurements in nano seconds not micro, and do specific buffering and cut-through switching and specific packet processing.

And before some gamer goes asking for what switch they want to shave off a couple of micro seconds off their ping time to game server 2000 miles away on the internet.. No it won't and you couldn't afford them anyway ;)

coffeecup25

@johnpoz

Yet I see what I see and I don't need 1000 words to justify it. All I changed was the switch and I noticed a little more oomph that wasn't there before all over my network, but especially on my Roku Ultra 2024. It went from fast to 'did you notice that' after the switch change.

And I am very sure I made it clear it was not light speed vs snail speed.

The rest including web pages was in the eye of the beholder - and I saw a snap that wasn't there before. Sometimes you just have to believe your eyes. Others in the house didn't see any difference on web pages. I did. Not everywhere but still there. I suspect it may be a biological perception issue. Some people can draw well, some are athletes. I have fast perception like many gamers seem to. Just a hunch.

coffeecup25

@scottjh1

I bought a TP-Link unmanaged 24 port basic switch and plugged it in. No configuration or skills needed.

Also, your pings are on your LAN. (192.168.1.51). So what? 2 tin cans connected by a string could do almost as well.

How about something that requires a lot of back and forth connections out into the world to complete a basic transmission? Like normal internet. The router is obviously doing a lot of that but switch is still involved.

dennypage

@coffeecup25 It's impossible to know after the fact, but one of the more likely items would be a link negotiation error, either duplex or flow control, leading to a small packet loss when busting through the switch. Relatively easy to diagnose when live, but impossible to know now.

Regardless, whatever you problem was it seems to have been resolved. Going back to your original question, my recommendation would be to recycle the decade old switches. This is likely justified on power consumption alone.

coffeecup25

@dennypage

Power consumption is a good thing to explore. In general terms, how have things changed? I paid about $13 for a new 5 port switch. The 8 port was $17. Very cheap. What kind savings are we looking at, more or less? It would be a false economy to donate it and see one sold for $10 if the excess power consumption would eat that up in a year or two..

Switches are low powered items. But, on the other hand, I saw a decent drop in my electric bill after I cut the cord and tossed out my Tivo equipment a few years ago.

dennypage

@coffeecup25 said in Toss Out or Donate Old Switches:

Power consumption is a good thing to explore. In general terms, how have things changed?

In general, power consumption (and generated heat) of most electronics has dropped precipitously over the past 15 years. That said, you would need to explore the specifics of your own devices to know the potential benefits to you of replacing them.

Patch

@coffeecup25 sorry to cause offence.
To answer your op,
if you can really notice a difference in user response time from swapping out a switch that indicates a switch fault (hardware, configuration, cables). If you have eliminated configuration, cables and device reset that leaves faulty switch hardware so toss it.

coffeecup25

@EdisonTaylor

Thank you for your thoughtful reply. Even if the switches use more power than today's models, they still use about as much electricity as a night light. So, power consumption may be higher but is realistically not an issue.

Yes, they all work very well and look new. I will donate them to ReStore locally or a thrift shop. People can decide for themselves if they are priced right. The improvement in responsiveness appears to be in the eye of the beholder.