1. This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn More.
Dismiss Notice

Welcome To SNBForums

SNBForums is a community for anyone who wants to learn about or discuss the latest in wireless routers, network storage and the ins and outs of building and maintaining a small network.

If you'd like to post a question, simply register and have at it!

While you're at it, please check out SmallNetBuilder for product reviews and our famous Router Charts, Ranker and plenty more!

How Many SSIDs (WiFi Names) is Too Many?

Discussion in 'General Wireless Discussion' started by Klueless, Jun 2, 2019.

?
  1. > 1

    28.6%
  2. > 3

    28.6%
  3. > 6

    14.3%
  4. > 9

    0 vote(s)
    0.0%
  5. Need 'em - Use 'em!

    28.6%
Multiple votes are allowed.
  1. Klueless

    Klueless Very Senior Member

    Joined:
    Jan 1, 2016
    Messages:
    898
    Location:
    Rochester, NY
    How Many SSIDs (WiFi Names) are Too Many?

    Up front I’m going to say if you need ‘em – use ‘em. Let’s take it further; if you want ‘em – use ‘em.

    The conversation often comes up as a side bar in other threads. There seems to be two camps; use as many as you need/want or use as few as possible. A discussion ensues and, a week later, it comes up again as a sidebar in yet another thread. So, I thought, perhaps this conversation deserves it’s very own thread?

    For some reason I became intrigued (possibly because I get teased for using too many SSIDs). I did some reading. Some material I understood, some I didn’t and some sources conflicted with other sources. I’ll spell out my assumptions to make it easier to agree, disagree and/or correct me and maybe we’ll put this issue to bed.


    A “WiFi Name” is called a SSID (Service Set Identifier). WiFi (wireless routers, wireless access points) must advertise their services so a “Beacon” is sent for each and every SSID. A beacon is a data packet and, as such, takes time and space. This is “overhead”.

    Some of us use as few SSIDs as possible to reduce overhead. Others (me) take the Alfred E. Neuman (Mad Magazine) approach of – “What, me worry?”

    Let’s start with one of my typical configurations:

    · “Whole House”; both 2.4 GHz and 5 GHz (family password)
    · “Faster”; 5 GHz (family password)
    · “Further”; 2.4 GHz (family password)
    · “Guest”; both 2.4 GHz and 5 GHz (guest password)

    In a manner of speaking that is six SSIDs but since 2.4 GHz and 5 GHz are separate collision domains I’ve really the overhead of only three SSIDs on 2.4 GHz.


    Now typically a SSID is advertised every 100 ms so that’s ten packets a second. I have three SSIDs so that’s thirty packets per second. Now let’s say I add an Access Point and call it “Garage”. That access point is going to advertise “Garage” so now I’ve four SSIDs and the overhead is forty packets per second (PPS). So I change “Garage” to “Whole House” to reduce overhead? Well that AP is still going to advertise what it knows (“Whole House”) so I’m still at 40 PPS even though I’ve only “three” SSIDs.

    2.4 GHz has eleven (1 – 11) channels available. Channels overlap so many say we only have three effective channels; 1, 6 and 11. My router and AP both use channel 6 thus 40 PPS. I change my AP to channel 1 and channel 6 is back down to 30 PPS (but, more importantly, my garage traffic is isolated from my house traffic).

    Now my neighbor is also using channel 6 and he also has three SSIDs. Between the two of us we’re advertising six SSIDs at 60 PPS. Hopefully I can change channels (11?) and I’m back down to three SSIDs and 30 PPS. But, again and more importantly, our data isn’t competing with each other for limited bandwidth.


    Now a credible opponent to multiple SSIDs is “Andrew” from “Revolution Wi-Fi”. He created a graphic representation to explain to engineers and customers why multiple SSIDs are bad and he is the most oft quoted resource by other opponents of multiple SSIDs. (You can read all about it right here ==> https://revolutionwifi.blogspot.com/p/ssid-overhead-calculator.html ) His front page conclusion is 3.22% overhead per SSID. No wonder opponents of multiple SSIDs feel the way they do. That’s a whopping 10% hit for just three SSIDs! (Click on image to enlarge.)

    Default.jpg

    But his front page conclusion is based on some ancient implementation of 802.11 B at just 1 Mbps. The lowest standard I’ve ever seen in the US is 802.11 B at 11 Mbps. Using Andrew’s own spread sheet against him that comes in at a much more reasonable 0.43% overhead per SSID. See all the "green" (acceptable) boxes now.

    B 11 Mbps.jpg

    Let’s take it a step further; Andrew’s spread sheet shows 802.11 G at 54 Mbps coming in at just 0.11% per SSID.

    G 54 Mbps.jpg

    Let’s take a leap of faith here and speculate that N and AC will do even better.

    In another thread @Rumboogy actually ran benchmarks with one, two, three and four SSIDs using AC over 5 GHz. He saw no degradation much less anything in the neighborhood of 10 to 13%.


    rumboogy test.JPG

    So let’s look at that. He benched just shy of 60 MBps (about 600 Mbps) with just one SSID and with four SSIDs. Assuming worse case (packet size = MTU = 1,500 bytes) that’s about 400,000 packets. 40 packets of overhead divided by 400,000 yields about 0.01%. No wonder he didn’t see any degradation.

    0.01% actually sounds reasonable to me. Round numbers AC runs over ten times faster than G thus beacons are a tenth the overhead which is what we just showed. And it might be even lower if you consider that I used @Rumboogy actual test speeds rather than the theoretical of 866.7 Mbps, I used maximum packet size and I didn’t even consider the positive benefit of multiple streams.

    Now granted the 2.4 GHz side has a more overhead but worse case of 0.43% is entirely acceptable. Running G or N will give us progressively better results. Eliminating backwards compatibility with B/G and running "N only" will achieve even better results.

    In conclusion I don’t see any reason that each and every Walton (from “Walton’s” TV show) can’t have their very own SSID.

     
    Last edited: Jun 2, 2019
  2. L&LD

    L&LD Part of the Furniture

    Joined:
    Dec 9, 2013
    Messages:
    12,991
    Just would like to point out a few things, besides that 60MB/s = 480Mbps, and not 600Mbps. :)
    1. Unless you are testing in a completely controllable environment, you can't really test the spreadsheets predictions because you can't get to 'zero' neighboring AP's, if there are already AP's around. :)
    2. When devices are inactive, I regularly see 1Mbps connection speeds in the System Log/Wireless Log page. Sometimes for transmitting, sometimes for receiving but usually for both, when not actively communicating on the network. :)
    3. With 20+ AP's surrounding our routers even in suburbia, let alone in apartment/condo style dwellings, the degradations are real. :)
    4. It doesn't matter what our (low powered) clients pick up in the WiFi environment, what matters is what the router picks up. I also believe that the router is not showing all possible connections either, but as conditions fluctuate over the course of the day, those other AP's will affect the performance of our router too. :)
    5. Also, the router will still be affected by non-WiFi interference, but won't list that anywhere for us to 'see' it too. :)
    6. The best settings for most routers is the Auto Wireless Mode. Significant degradations become apparent for some if other Modes are chosen, even if they are theoretically superior. :)
    7. WiFi, being a shared medium, is almost never only about the variables we directly control. :)
    8. With all the above said, for a single router, the main SSID and a guest SSID for each radio are more than enough to make for a solid WiFi network. :)
    9. For the main router and additional AP's, a single SSID for each radio is all that is practical (guest networks just don't work properly with an AP).
    10. Anything additional will introduce more overhead, resulting in a less responsive network, even if it isn't apparent in simple testing scenarios that only account for maximum throughput. :
    To expand on point one above, the overhead is not just in maximum throughput, but I think more importantly in the responsiveness of the network too. when a router has to negotiate with just itself or with a few dozen other AP's and SSID's it's juggling, it makes a real difference.

    Management frames are always sent at a legacy data rate using either 802.11b or 802.11g in 2.4 GHz, and 802.11a in 5 GHz.

    And don't forget about adjacent channel interference for those neighbors that use the in-between channels and wreak havoc for all. :)

    http://www.revolutionwifi.net/revolutionwifi/2014/08/80211ac-adjacent-channel-interference.html

    I loved the Waltons too, but I would never give them each their own SSID. :)
     
  3. Klueless

    Klueless Very Senior Member

    Joined:
    Jan 1, 2016
    Messages:
    898
    Location:
    Rochester, NY
    Wow, ten objections. I'm flattered that you took the time but, more importantly, I always enjoy talking with you. Of course I'm not going to address them all, but, maybe a few ; -)
    60 MB was a measure of payload. Packets have overhead, headers and trailers and such. Mbps is a measurement of how many bits go over the wire which counts payload as well as packet overhead. So I simply multiply by 10 rather than 8. It keeps the math easier while factoring in some packet overhead. In this case it also made my numbers more pessimistic, e.g., it worked against my case.
    I agree. I just called out the numbers for illustrative purposes.
    Juggling? My understanding was that a wireless router or AP simply broadcasts its SSIDs and it's up to the client to do what it will.
    Even better. Even with "interference" he didn't see overhead or variability.
    Ouch. That's a whole conversation in itself. First off I don't believe the router cares about neighboring SSIDs, its job is to simply advertise what it's offering.

    It's the client who must process all those offerings. Perhaps is good he's "under powered" as he sees less clutter/confusion.

    But it's still an interesting point as when the "higher powered" router does his "carrier sense" he's more likely to see a busy. In any case I don't see this as a "SSID issue"; I see it as a contention issue.

    But, in general, I'm happy with my "confusions" as I took the data Andrew used in his case against SSIDs and used it to make using multiple SSIDs a whole lot less terrifying. At least I thought so ...

    More later after my beer and cigar break ...

    Until then I still think "Mary Ellen" deserves her very own SSID : -)
     
    Last edited: Jun 3, 2019
  4. Rumboogy

    Rumboogy Occasional Visitor

    Joined:
    Jun 21, 2017
    Messages:
    41
    Instead of debating the theoretical someone could just gather some more data and share it. If a test could be done in an environment with a fair amount of congestion then this would probably be a useful data point for others.
     
  5. Klueless

    Klueless Very Senior Member

    Joined:
    Jan 1, 2016
    Messages:
    898
    Location:
    Rochester, NY
    • Ouch <lol>
    • I think you've already done far better than me! ("Me" being the jackass who teased you that your Internet "pinch point" of 300 Mbps might be compromising your tests : -)
    • None of my sites really lend themselves to testing which ... is one of the reasons I run with a couple extra SSIDs. Then, if a client is misbehaving, I can often effect a fix without a lot of testing, reconfiguration and disruption.
      • Although ... that need does seem to be waning as we update our older equipment.
     
    Last edited: Jun 4, 2019
  6. dosborne

    dosborne Very Senior Member

    Joined:
    May 11, 2019
    Messages:
    594
    Location:
    /dev/null
    Personally, I don't think there is any one answer (at least not from the list presented). It all depends on how many devices you are connecting and what function these devices perform and what security these devices require and what performance these devices require.

    Installing a Smart Power Plug that gets turned on or off once a month is different form installing a router and that is different from allowing a guest to bring their unsecure device into my home and connect using my internet and that is different from my $5000 gaming system (although I don't have one, you get the idea).
     
  7. Klueless

    Klueless Very Senior Member

    Joined:
    Jan 1, 2016
    Messages:
    898
    Location:
    Rochester, NY
    Yeah, I regretted that "poll" as soon as I hit "submit" and, worse, I didn't see a way to change or delete it : -(
    You are touching upon an important point. Risk / Reward. Value. Value = Benefits less Costs.

    You touch upon some of the benefits, e.g., a "guest" SSID that keeps them away from your privates. At one place I had a SSID for company devices (with a "secret" password) and another SSID ("BYOD") for devices that employees brought to work (for which we shared the password).

    It actually came in handy a few times. If, for example, we had problems (we only had a 7 x <1 Mbps service back then) and no time to trouble shoot we could simply shutdown "guest". If that didn't fix it we'd shutdown "BYOD" and get back to work.

    (Turned out it was high utilization. New iPhones that wanted to "replicate" over our dirt (s)low speed net. Boss' kids bringing their gaming machines to work. The longer term fix was properly setting up QoS.)​

    "Revolution" did a nice job explaining "costs" (overhead).

    So if there's no benefit why incur the cost? But if there is a benefit (real or perceived) then go for it. Do understand there is a "cost" but it's not near as draconian as some have been led to believe.
     
    Last edited: Jun 4, 2019
  8. Klueless

    Klueless Very Senior Member

    Joined:
    Jan 1, 2016
    Messages:
    898
    Location:
    Rochester, NY
    Now that got me to thinking. Is it just checking to see if it really has to bother talking at low speeds?

    I was pretty sure "Revolution" had accounted for that but ... who knew for sure? Even though he allowed for a pick list to select different scenarios he locked the spread sheet thus I couldn't see his formulas.

    So I wrote my own spread sheet. What I did was calculate how many ms it would take to send the beacons at 1 Mbps, then I calculated how many data bits I could send in the remaining time at 1, 11 B & 54 G. Then I calculated overhead as beacon bits divided by data bits. I didn't get the same numbers as "Revolution" (mine were a little lower) but, in similar fashion, the important part was that I saw overhead go down even with beacon bits being calculated at 1 Mbps. (Click image to view.)

    KluelessOverHeadCalc.jpg

    [Edit]: I reread my spread sheet and I think I'm full for crap. Using "time" for my calculation doesn't account for the extra data I could have sent during the time I used up sending beacons at the lower speed. I'm now wondering if overhead remains a constant if you're anchored at a lower speed but overhead matters less at higher speeds.

    So now I'm left wondering if "1 Mbps" even happens if I turn off "B/G" compatibility and/or run at "N only"? Or if anyone really bothers to support "A" at 5 GHz? Or how much any of that will really matter as we update our old computers, printers and smart phones?
     
    Last edited: Jun 5, 2019
    Marin and L&LD like this.
  9. sfx2000

    sfx2000 Part of the Furniture

    Joined:
    Aug 11, 2011
    Messages:
    14,340
    Location:
    San Diego, CA
    Might want to review the article on the main site...

    https://www.smallnetbuilder.com/bas...751-snb-answer-guy-how-many-ssids-is-too-many

    Single SSID is a very valid approach, and to the end-users, makes things fairly seamless.

    Maybe one additional SSID for the guest network if needed.

    Most of the issues with sticky clients relate to having multiple AP's, and having them too close together, so the client doesn't get a trigger to scan for a better AP.
     
  10. dosborne

    dosborne Very Senior Member

    Joined:
    May 11, 2019
    Messages:
    594
    Location:
    /dev/null
    Personally, if the question/poll was reworded to "What is the minimum number of SSIDs you should have?", my answer would be:

    +1 for 2.4GHz connectivity (if you have these devices)
    +1 for 5GHz connectivity (if you have these devices)
    +1 for (2.4GHz) Guest connectivity (if you have guest devices)

    :)
     
  11. Klueless

    Klueless Very Senior Member

    Joined:
    Jan 1, 2016
    Messages:
    898
    Location:
    Rochester, NY
  12. Klueless

    Klueless Very Senior Member

    Joined:
    Jan 1, 2016
    Messages:
    898
    Location:
    Rochester, NY
    Yeah, I still regret hitting "submit" on that poll : -(

    I guess I could make do with just two;
    • One for "guest" largely because I can turn off access to my Intranet to keep 'em away from our privates.
    • And one more for both bands because it's simple and all "automagic".
    But often I do add two more; one for 2.4 GHz and another for 5 GHz because I still run into too many devices that make bad decisions. At the risk of repeating myself I often call them "faster" and "further" because it's kind of self documenting and the user, if he's having problems, can often figure out how to fix it themselves.

    I'll give a couple examples;
    • At one site the boss was only about 25' feet and two walls from the router. Worked fine but would fail enough to make him grumpy. (He often had many people in his office, standing in front of the wall and/or sitting in front of his desk between the computer and router.) On a whim I set him to "further" at 2.4 GHz. Problem went away. True, he didn't bench at 300 Mbps anymore but he didn't go down any more either,
    • One staff member was about 30' and one wall from the router. It also worked fine but would fail enough to make her grumpy. (We'd often bring a large SUV into the showroom which would sit between her and router.) Set her to "further" and problem went away.
    • At home we have kids, grandkids, Rokus, smart TVs and gaming machines. Had random problems with a couple devices (TVs) so we manually set "faster" or "further" as indicated. Had problems with one of the grandkids (gaming machines), I don't think he was really having a problem (with our net) but I showed him "faster" and "further" and he was happy if, for no other reason, he felt he was in control.
    OK, I've bored you long enough!
     
    Last edited: Jun 6, 2019
  13. Klueless

    Klueless Very Senior Member

    Joined:
    Jan 1, 2016
    Messages:
    898
    Location:
    Rochester, NY
    Based of feedback, additional reading and some more calculations I'd like to temper my opening statement from "... If you need ‘em – use ‘em. In fact let’s take it further; if you want ‘em – use ‘em!" to simply ... "If you need ‘em – use ‘em".

    Several reasons;
    • Being a good neighbor
        • I pretty much stick to 20 on 2.4 GHz and 40 on 5 GHz for that reason and because channels are a scarce resource so I shouldn't pig them all up.
      • I stick with channels 1, 6 or 11 ( such that running channel 3 won't compromise my neighbor's use of channels 1 and 6).
      • I do not amplify transmit.
      • So I shouldn't broadcast needless SSID advertisements until I know more. That said if there is a benefit to me I'll use them until I know more.
    • I'm still trying to understand SSID overhead. I'm not going to be shy about using multiple SSIDs but I'm not going to go crazy either until I know more. I will lean more towards benefits and further from "just in case".
    I've reworked my spread sheet and my numbers for B @ 1 Mbps now match Andrew of Revolution's so I'm pretty proud about that. My numbers for B @ 11 and G @ 54 Mbps are lower than Andrew's which suggest that his numbers allow for backwards compatibility? If so the cautions of 2014 may become a nit in the near future?

    One of my routers allowed me to turn of B/G compatibility and run "N only". I wonder if that lowered my overhead (e.g., eliminated the overhead of backward compatibility)?

    I'm still thinking about the @Rumboogy tests where he couldn't detect any difference in overhead between 1 and 4 SSIDs. It just hit me; he ran AC @ 5 GHz. Worse case for 5 GHz is only 0.56% so for 4 SSIDs (coupled with the fact he was running AC and not A) he's going to clock in at less than 2% overhead total. That and although he was benching an awesome 600 Mbps he was still a couple hundred away from theoretical so there was still plenty of extra bandwidth for beacons?

    Are we having fun yet?
     
    Last edited: Jun 5, 2019
    L&LD likes this.