Canadian TV, Computing and Home Theatre Forums banner

Is there a performance hit with too many switches?

6K views 38 replies 6 participants last post by  MyMedia08 
#1 ·
Here is the context....

I have an AP in the den, which is the main area re wifi usage. It's currently a flakey 8 port DIR 632 ; it's requires a reboot frequently.. Would love to replace it with a solid unit ..thinking something like the Netgear Nighthawk R7000 (about $180)

But I currently take advantage of the DIR 632's 8 ports, as I need to connect the Ethernet wired source, plus a roku box, pc, AVR, and blu ray ( ie 5 cables), so if I replace it with the Nighthawk, I'd have to add a gigabit switch.

Normally not a problem...but because of some obstacles in the house, and finished basement, the path to that AP already has two switches : Hitron CGN3 (office) -> gigabit switch (office) -> gigabit switch(basement) -> 8 port AP (den).

So if I go with a Netgear nighthawk, I would need to add yet another switch .... Is this a concern or non-issue??
 
#3 ·
ok thanks. Good to know given my choices are limited.

I'll take the opportunity to redeploy and reconfig what I can. Current plan is shown below re a future Nighthawk for the den. Wired Ethernet to den is only available via the basement.

Question - Are there some simple LAN benchmark tools (kindda like a SpeedTest for networks) that can help me get a baseline and spot performance roadblocks?

 
#5 ·
One suggestion I'd make is to distribute everything from the office switch, rather than the Hiltron router. If you run out of ports on the switch, move low performance stuff to the Hiltron. The idea is to minimize the number of hops when practical.
 
#6 ·
Thanks ! Hmmm, interesting, and I certainly have room on the office switch to move PC1 and PC2 there.

I may be misunderstanding something.... but right now PC1 and PC2 are direct to the Hitron, no hops
(ie CGN3 - > PC1)​
..so why would it be better to introduce a switch in between?
(ie CGN3 -> switch -> PC1)​
 
#7 ·
Curious. How big is your house?

Three wireless access points and a wireless router seem really excessive. Also you seem to have a lot of switches.

In my house, my wired router connects directly to two wireless routers, my Voip phone and my main switch. (this gives direct access to the internet for these devices)

Inside the house all wired connections go through the main switch (so internal traffic never reaches the wired router) and the wireless devices goes through one of the two wired routers since these devices are typically communicating with the internet and not internal devices.
 
#9 ·
Curious. How big is your house?

Three wireless access points and a wireless router seem really excessive. Also you seem to have a lot of switches..
Guilty with explanation...but wireless is off on the router....and I don't think I have to many options given the constraints I have (house is 3000 sf but layout is a little unconventional, lots of obstacles, and a finished basement.

SWITCHES - Re lots of switches, I have lots of hard wired devices..so use of other switches (or use of router ports) in various rooms are simply to connect many hard wired devices. No choice there. And the only path to the den (where I really want the wi-fi) is via a drop in the basement. So I must piggy pack on the existing switch there, to serve the hard-wired needs in the den.

WIRELESS - As to the multiple APs,, that was by design. When I first got on the wireless bandwagon, I went through many wi-fi routers, and none were 'strong' enough to service the whole house. It then occurred to me that as I was wired up in many rooms. all I really needed were some local APs in the right places, so for $20 each, I bought 3 cheapo blue Trendnets, and problem was solved.

MAIN ROUTER - When I later wanted to upgrade my main router, I was simply looking for something reliable and stable. Didn't need wireless, but at that point it was hard to find any without wireless, so I ended up with the DIR-655 (wireless off) and retained the 3 cheapo APs for the rest of the house.

UPGRADES - Fast forward to today. I just upgraded to the Hybrid 60 with the Hitron CGN3. Again wireless if OFF there so cant count it as wireless router :). The 655 is simply being redeployed as an AP for coverage in the office for now. I could remove it as I only rarely need Wi-Fi in that room ( a jut out from the house with concrete on three sides,) so an AP in this room works well. But I suppose it can go if I don't use it. The other cheapo APs are dated and I want to replace them, starting with the den.​

I agree that if the Netgear Nighthawk for the den will prove "strong" enough I may not need the other APs, so they may disappear over time...but I will still need local switches to connect hard wired devices in those rooms.

Do you see any other opportunities to reduce switches/routers given this setup?
 
#8 ·
Extra note: I used Wifi Analyzer for my Android smartphone to determine wireless strength throughout the house. This can help establishing if you really need so many wireless access points Great tool that I believe was free or just a buck or so.
 
#30 ·
My 2 cents worth on the NightHawk. Mine works extremely well with way better coverage than my previous Cisco model...... Single strength was so good in fact I actually reduced it from the default setting to minimize EMF (WAF).....
Is there an app for the ipad to show Wi-Fi signal strength?
 
#10 ·
Random thoughts: From your diagram, I don't get why you have both a switch and AP in your office when one should suffice. Also I don't get why you would buy the Netgear AC Router and then turn it into an AP for really just one room. You don't seem to have AC devices so I'd recommend a cheaper but good quality N router and upgrade to AC when the prices come down. I'd also get the Wifi analyzer out and determine if you really need so many AP's. Don't forget AP's can be on the same SSID but should be operating under different channels or you may just end up making your Wifi worse.

Good Luck!
 
#11 ·
Random thoughts: From your diagram, I don't get why you have both a switch and AP in your office when one should suffice.
Thanks for the review. Re office, the AP was an after thought. You are correct, I could either shelve it (give up wifi in office) and just keep the switch, or shelve the switch and just use the DIR 655 as an AP/switch. I am still curious about JamesK' suggestion to distribute everything from the office switch, rather than the Hitron router.

Also I don't get why you would buy the Netgear AC Router and then turn it into an AP for really just one room.
Its not just for one room. Ideally it will serve many rooms, outside etc. Recall earlier the critique of too many APs and just have one ! Hopefully the new Nighthawk will eliminate the need for some of the older APs.

You don't seem to have AC devices so I'd recommend a cheaper but good quality N router and upgrade to AC when the prices come down.
Several reasons. I think if I buy two devices (one now, upgrade later) , it will end up costing the same anyway. And I will still benefit from the new router even without AC devices (its quantum leaps better that what I have now.) I may have AC devices in future as I replace older hardware. Finally, buying an N now and an upgrade alter leaves me with yet more hardware to redeploy down the road. I saw no harm buying a solid unit now.

I'd also get the Wifi analyzer out and determine if you really need so many AP's. .
I did that when I first started. Walked around with a laptop and connected to various APs and concluded the local AP was need and worked best. At at $20, it was a no brainer. But yes, I will repeat that exercise when I get the Netgear Nighthawk to see if I can eliminate some APs.

Don't forget AP's can be on the same SSID but should be operating under different channels or you may just end up making your Wifi worse..
Very true. Initial setup things re Ch 1, 6, 11 etc, but I should re-confirm what I am using now.

Thanks again for the ideas.
 
#12 ·
.so why would it be better to introduce a switch in between?
Compared to a gigabit network, a cable modem is low performance. The latency of the switch is negligible when compared with that of the "net". Of course, the best configuration depends on how you use your network. If you mainly use the Internet and do little between devices, then connecting directly to the Hiltron is fine.

By comparison, a larger enterprise network has multiple layers. There's the core, (sometimes) distribution and access layers. The users are connected to the access layers, but everything generally tries to have the fewest hops to the core. The Internet connection might be connected to the core, but may be further down, depending on needs. In your case, you might consider that office switch to be your "core".

I am still curious about JamesK' suggestion to distribute everything from the office switch, rather than the Hitron router.
The Hiltron doesn't have enough ports to meet local needs, but the switch does, so the switch becomes the core.
 
#15 ·
Compared to a gigabit network, a cable modem is low performance. The latency of the switch is negligible when compared with that of the "net". Of course, the best configuration depends on how you use your network. If you mainly use the Internet and do little between devices, then connecting directly to the Hiltron is fine....... In your case, you might consider that office switch to be your "core"...
Okay thanks. I think I understand a bit better....ie that if there is a lot of traffic between devices (which there is re my frequent backups, replications etc) , its better if they all on the same core switch, rather than have some on the switch on some on the router. I'll give it shot.

As to the lower performance cable modem compared to gigabit networks, on one hand I cant do much there ( ie cable modem must be connected to the core switch)....but are you also saying this is an example of where bridge mode is better than gateway? ie let the router functions be handled by a higher performance device?
 
#13 ·
Also you seem to have a lot of switches
That may be determined by cabling requirements. For example, it may be easier to run a single cable back from a remote switch than individual cables for each device. In my condo, I have a single Ethernet cable running to my living room, where I currently have 3 devices connected to a switch. That cable was pulled in by Rogers, when they ran in the coax for my cable modem, to my "office" where my computers are located. It would be a lot of work to run in more cable, so I put a switch in.

I used Wifi Analyzer for my Android smartphone
There is also inSSIDer, which covers 5 GHz in addition to 2.4. I don't believe WiFi Analyzer covers 5 GHz.

Don't forget AP's can be on the same SSID but should be operating under different channels or you may just end up making your Wifi worse.
Don't forget about the neighbours. The more channels you use, the more likely you are to experience interference with them. Also, if running N, on 2.4 GHz, set to max data rate, you're occupying 2 of the 3 non-overlapping channels.
 
#16 ·
^^^^
No, I was just pointing out what to consider when designing your network. If most of your traffic is on the Internet, then make the Hiltron your core. If not, then that office switch. If the Hiltron were in bridge mode, then you'd need another router and then you're back to the same situation.
 
#18 ·
One thing to keep in mind is that the bottlenecks are the cables, not the switches and routers. (That assumes the switches and routers are all 1Gb models. Slower switches will be a bottleneck.) Most good switches will handle 1Gb/s between multiple ports, maintaining full speed on all connections, but a cable is limited to 1Gb at most. That's important if multiple devices need full 1Gb speed back to a central device such as a server. If that is the case, it's best to employ a star configuration with all remote switches and high demand devices hard wired back to a central switch near the server. (Some servers employ multiple ethernet ports to handle multi-Gb loads.) Cascaded switches may cause bottlenecks on the busiest cable link. It's also important to use high quality cables, such as CAT6 or better. I'm not saying cascaded switches are always bad but they require a load analysis on the cable links.
 
#19 ·
Some of my wiring is fully accessible, but much is buried in the walls and can't be upgraded. But I am pretty sure I used Cat5e at the time, which I think is gigabit rated.

But how do you measure potential bottlenecks ? Ie what's a decent network performance monitoring tool for home use ? I tried Qcheck and Netstress, but a they don't seem to be all that convenient to use.
 
#20 ·
The simplest way is to copy a large file and look at the transfer speed. If it's over 500Mb/s then the network is OK. Anything over 250Mb/s is usually adequate. Most applications will not require over 50Mb/s anyway. If the speed is under 50Mb/s then troubleshooting is required. That can often be traced to an individual device, such as the PC itself.
 
#21 ·
Honestly mymedia, I think you might be overanalyzing this.

Exdilbert's first comments are excellent in a corporate environment but I'm not sure you should worry too much in your home where generally you will likely never saturate your network.Certainly test a file transfer but an extra switch really shouldn't materially impact the transfer.

I would simply try to reduce the number of switches and AP points so there are fewer hops and keep the critical devices (a game console or IP phone) connected directly to the first switch to reduce any latency.Most routers prioritize Voip or game traffic automatically so it may not even be an issue.
 
#23 · (Edited)
Honestly mymedia, I think you might be overanalyzing this...
Only partial true in that I was looking for some quick and easy off the shelf tools. But I think it fair game to determine if bottles exist.

I now see the easiest way to address my concerns is with some simple large file transfer tests.

I whipped up some small bat files/spreadsheet to faciltiate that process, to copy a 500MB file to PC01 from each of my other pc's. All hardwired. Indeed very worthwhile and I have major bottlenecks on PC03, PC04 and PC06. These bottlenecks look like more than simple the result of an extra switch or longer cables, so I need to dig a bit deeper, perhaps to NIC settings , flakey boxes ( DIR 632??), etc.


PC01 TO PC02 (via core Switch) - OK
============
Date test completed............ :2014-06-29
Time test initiated............ :13:22:40
Time test completed............ :13:22:45
Actual DOWNLOAD TIME in seconds :4.48 seconds
Actual FILESIZE in bytes....... :536,870,912 bytes
Computed FILESIZE in bits...... :4,294,967,296 bits
Computed BITS/SEC.............. :958,698,057.14 bps
Computed Mbps.................. :914.29 Mbps


PC01 TO PC03 (via core Switch and longer cable.). EXPLAINED. This is old XP and likely a non gigabit NIC , to be confirmed.

============
Date test completed............ :2014-06-29
Time test initiated............ :13:22:45
Time test completed............ :13:23:32
Actual DOWNLOAD TIME in seconds :46.66 seconds
Actual FILESIZE in bytes....... :536,870,912 bytes
Computed FILESIZE in bits...... :4,294,967,296 bits
Computed BITS/SEC.............. :92,048,163.22 bps
Computed Mbps.................. :87.78 Mbps

PC01 TO PC04 ( via core Switch and DIR 632 port and longer cable ). NOT GOOD.
============
Date test completed............ :2014-06-29
Time test initiated............ :13:24:16
Time test completed............ :13:25:02
Actual DOWNLOAD TIME in seconds :45.88 seconds
Actual FILESIZE in bytes....... :536,870,912 bytes
Computed FILESIZE in bits...... :4,294,967,296 bits
Computed BITS/SEC.............. :93,613,062.25 bps
Computed Mbps.................. :89.28 Mbps

PC01 TO PC05 (via core Switch and longer cable). OK
============
Date test completed............ :2014-06-29
Time test initiated............ :13:25:02
Time test completed............ :13:25:07
Actual DOWNLOAD TIME in seconds :5.53 seconds
Actual FILESIZE in bytes....... :536,870,912 bytes
Computed FILESIZE in bits...... :4,294,967,296 bits
Computed BITS/SEC.............. :776,666,780.47 bps
Computed Mbps.................. :740.69 Mbps

PC01 TO PC06 (via core Switch, DIR 632 switch, longer cable and temp AP switch). NOT GOOD
============
Date test completed............ :2014-06-29
Time test initiated............ :13:27:11
Time test completed............ :13:27:56
Actual DOWNLOAD TIME in seconds :45.34 seconds
Actual FILESIZE in bytes....... :536,870,912 bytes
Computed FILESIZE in bits...... :4,294,967,296 bits
Computed BITS/SEC.............. :94,727,995.06 bps
Computed Mbps.................. :90.34 Mbps


 
#24 ·
I'm confused. Based on your equipment diagram, those numbers are close to what I would expect. You can't expect Gigabit transfers through 10/100 ports.
 
#25 ·
Ooops my bad. Of course. Too much redeploying of hardware. Could have sworn the 632 had gigabit ports! . So that explains the PC04 and PC06 results.

Will repeat all tests once the Nighthawk arrives and will replace the 632 with my spare DIR 655 (which has gigabit ports).
 
#26 ·
There is 10 Gb over copper and much higher over fibre
Maybe I should have qualified that with, "using gigabit switches" but I thought it was obvious in context.

MyMedia08, definitely use Gb switches for maximum speed. 100Mb devices might be Ok for a wireless B/G access point or with 100Mb devices but they will be a bottleneck anywhere else. One thing I maybe should have mentioned is that Windows Explorer reports rates in MB/s or KB/s. Multiply by 10 to get the approximate Mb/s or Kb/s.
 
#27 ·
Cat 5e is gigabit rated, whereas Cat6 is 10Gbit rated. If you're operating a gigabit network, Cat 5e should be fine. However, Cat 5 cable is only rated at 100 Mbps.

Your cabling is unlikely to be the bottleneck unless it's (a) less than Cat 5e, or (b) kinked or bent. The main bottleneck is likely to be those 100 Mbps switches or NICs, followed by the hard drive read/write speeds.

As far as switches go, we installed some tiny 5-port D-Link GO-SW-5G (or GO-SW-8G for 8 ports) switches at work and they work beautifully, and they're dirt cheap. I'd recommend them as replacements for your DIR 632.
 
#28 · (Edited)
Pleased to report that all my prior bottlenecks are now addressed. The plentiful but required switches and APs (and longer cable paths) proved not to be an issue. Re LAN performance, I confirmed I am getting near gigabit transfer rates even to the PC at the end of the chain. Re internet performance, I am getting slightly better than the Hybrid Fibre 60/10 speeds on all devices , typically 77/11 Mbps, except for the ipad2 which is maxing out at around 40/11 Mbps).

The last of my old cheapo b/g routers, which served their purpose at the time for older machines, is now gone. I re-deployed the DIR632 for the only remaining non-gigabit guest pc. It will be turfed if and when that guest pc gets replaced. I redeployed the DIR-655 to the basement to serve as a g/b switch and local AP.

So all is now good.

The new Nighthawk is arriving today for use as my main AP to serve the heavy-use den and surrounding area. Will post to appropriate threads if I get into any issues there.

Again, thanks for all the discussion and input.

 
#29 ·
My 2 cents worth on the NightHawk. Mine works extremely well with way better coverage than my previous Cisco model. I flashed it with DD-WRT to give me some extra options. I was able to get coverage throughout my house without using a repeater or AP. Single strength was so good in fact I actually reduced it from the default setting to minimize EMF (WAF). BTW, I use Ethernet over power for hardwired connections and have 2 Apple TV's, 3 Roku's, my AVR, and 3 smart TV's connected with a few home computers.
 
#31 ·
I don't think there's anything similar to Wifi Analyzer on (non-jailbroken) iphones/ipads. A quick Google search indicated that these types of app were arbitrarily removed from the iTunes store a long time ago as a policy decision.

According to the developers of Network Analyzer, a network diagnostic tool for iOS 7, Apple restricts access to information such as SSID, signal strength, channel number, etc., preventing apps from accessing such information.
 
#32 · (Edited)
Its ironic how flawed the Apple policy is. In my search, I tried some free IOS apps. All showed a list of SSIDs (which you would think should be hidden) , but none showed an innocent , generic number like signal strength. Bizzare.
 
#34 ·
I think you have to pay for inSSIDer so I use Wifi viewere from nirsoft.net which is a free download for my laptop. May not be a extensive but it gives me what I need.
 
#35 · (Edited)
I don't have any android phones, but when my kids come over to house next, I'll have them download WiFi analyzer on their phone. My blackberry was another option, but the apps are too lame.

Will check out Wifi viewer from nirsoft.net , thanks.

I also stumbled on the free XIRRUS Wi-Fi inspector for windows. Downloaded it on my desktop ( which has a wifi adapter) and its quote nice. User guide is here.


 
This is an older thread, you may not receive a response, and could be reviving an old thread. Please consider creating a new thread.
Top