Article Discussion: How Many SSIDs Is Too Many?

[htismaqe]

@L&LD

I know we've discussed this before, but your methodology doesn't scale. I'm sure it's easy enough to explain separate SSIDs to one guy, or 10 guys, or maybe even 50.

But what if you're deploying wireless for hundreds, or even thousands, of users?

At some point, the network needs to be "set it and forget it". Most companies don't care about maximum network performance, they care about maximum profits. As long as network performance doesn't interfere with that, it doesn't need to be messed with.

 

[azazel1024]

I would liken this (a single ssid) to someone that does not care to have the most performance from their network and changing to another ssid if / when they move to another area is just too bothersome for them.

The networks that I setup for customers are usually not so big in area that the main routers signal drops completely. Rather, almost everywhere that you can go on the property your devices can 'see' all three or four routers / ap's. A single ssid in those situations is a recipe for disaster, ime.

Unique ssid's that are specifically understood to be used when in certain areas is far easier to manage, teach and learn than having people complain that they are connected, but still can't browse the 'net.

I am sure the opposite of the situation above occurs at times too. But my customers and I don't have the time to be testing if a single ssid will work when they can just point to the email from 'last Friday' to let people know what they need to do to stay connected. Just select the ssid (and same password) for the area you just moved into.

Err, well, since my gear switches pretty seamlessly and appropriate, there is basically no performance hit on my network at all. Actually probably better performance than trying to get my family/friends to switch manually. If I was only me and me alone and my devices in my house, yeah, I could probably manually switch bands and base stations between devices to coax the last erg of performance out of my wireless network. As it is, allowing the gear to select the appropriate band and basestation, which it almost always does, gives MASSIVELY better performance than trying to tell my wife "hey, if you go down to the basement, don't forget to switch networks. Oh hey, 7 year old son of mine, if you are going to take your tablet out on the deck, switch networks".

User managed connections only work with advanced users, otherwise performance suffers for everyone (because you have users connected at long distance, slowing everyone else down on that SSID/band).

 

[azazel1024]

@L&LD

I know we've discussed this before, but your methodology doesn't scale. I'm sure it's easy enough to explain separate SSIDs to one guy, or 10 guys, or maybe even 50.

But what if you're deploying wireless for hundreds, or even thousands, of users?

At some point, the network needs to be "set it and forget it". Most companies don't care about maximum network performance, they care about maximum profits. As long as network performance doesn't interfere with that, it doesn't need to be messed with.

Agreed. Separate SSIDs might work if you are talking a very small business or home where you might have a single AP with two bands, or maybe a pair of APs with a couple of bands. Maybe you are lucky that your clients are all reasonably proficient. My experience in life is that even most IT literate people are lazy. Or if not lazy, are annoyed (even if not openly) if they have to be changing their wifi connections. In situations where people don't have to roam around much and thus rarely have to change their connection, it might be viable...but then again, why would you need to worry about separate SSIDs anyway? The client is going to connect to the local one and physically and wirelessly be camped anyway.

WiFiVoIP is getting to be very common. At home, my wife would KILL me if anytime she took a call on her phone she had to change wireless networks walking around the house talking on the phone. In a business setting, probably worse. I know plenty of people in my office who absolutely walk around on their BBs (yeah, I know, quaint) taking calls, answering emails, etc. going to and from meetings. With buildings on the order of hundreds of thousands of square feet, having them stop, terminate a call, change wireless networks and recall, absolutely not acceptable. Heck, even in some smaller offices I have been in for training and that kind of thing, having to do the same even if you were only talking 4-6k sq-ft of office space just wouldn't work. And yes, I know most phones are seamless cell fall back for VoIP, but most of my office buildings don't have acceptable (in some cases ANY) wireless coverage. Doesn't help that we had a Verizon contract with Verizon nano cells (both business and personal cell was relayed through them), when we switched to AT&T, Verizon shutdown (but did not remove) the nano cells so that personal cells wouldn't work through them (and 80% of the builds are complete cell dead zones and AT&T has still not deployed nano cells even something like 3-4 years in to their contract).

If you don't have WiFiVoIP, it might be more tolerable. I'd bet if you asked anyone in an office environment though which way they'd prefer it "have to change your SSID every time you moved towards a new WAP, or have one SSID and let your device change on its own", I doubt you'd have anyone willing pick the former. Just because they can do it and will do it, doesn't mean they "know better" or it isn't a mild annoyance to them.

 

[L&LD]

It does seem I am the official opposition of this point of view.

Either way, it is usually not me that has to explain things to the end user. I simply make a word document with less than 20 words that the owner of the business can send to users as needed.

I should also clarify that when I do setup a wireless network, the goal is maximum performance because that is usually one of the reasons they called me in the first place. To get there, the main areas that need coverage are what are emphasized. The areas in between are not important (eg. the water cooler in the hallway) and I explicitly state that when I submit my proposal.

No one (the one's paying me) has complained about this approach yet. And the cost savings for 'full' coverage are significant. While the increased performance of their network (overall) is as obvious as night and day.

If I could concentrate on being the sys admin for a company, I would be able to ensure that all allowed devices worked seamlessly on a single ssid over multiple AP's and bands. But since if / when my customers call me they will get billed, my method up to now has offered them a working solution at minimal cost (my billing them) and time wasted (by me, the employees and my customer who is usually their boss).

Modems, routers and switches are setup so that rebooting in the proper order is as easy as '1, 2, 3'. This solves many, many issues (especially in static prone locations or seasons) and all that is needed are up to three power bars labeled as '1', '2' or '3' and a note on the wall to wait 5 minutes before resetting the next higher numbered one to 'on'.

Like I've mentioned before, I can see how a single ssid would be beneficial. But that seems to be a certain kind of utopia right now. In my experience with the many varied devices my customers and their employees depend on to keep that business running, a single ssid causes more issues than not at this point.

Maybe by the second or third generation of routers like the RT-AC3200 we will have a truly bulletproof single ssid protocol. But until then, I get paid to solve 1) performance issues and 2) setup a system that runs by itself (for the most part) and not setup a system which keeps me being called back indefinitely for the same issues (a 'make work' project for myself, as one customer told me).

 

[thiggins]

This is like watching people argue how many angels can dance on the head of a pin...

 

[htismaqe]

This is like watching people argue how many angels can dance on the head of a pin...

Well that answer is easy!

 

[sfx2000]

Maybe by the second or third generation of routers like the RT-AC3200 we will have a truly bulletproof single ssid protocol. But until then, I get paid to solve 1) performance issues and 2) setup a system that runs by itself (for the most part) and not setup a system which keeps me being called back indefinitely for the same issues (a 'make work' project for myself, as one customer told me).

Just consider that in the SME space, and large Enterprise/Carrier space - SSID's are commonly deployed as singleSSID - even there, with VLAN's deployed where AP's may use multiple SSID's (CorpGuest, CorpWiFi, CorpVOIP, for example), these are common across all AP's...

In the SNB/SOHO space, "Lazy Roaming" with SingleSSID generally works - you typically won't find directed handoff's, channel lists for AP discovery (although, I've seen some AP's do this, and when it is deployed, it's pretty cool if the clients also support it) -

L&LD has a good point - everyone's situation is different, and for some networks, SingleSSID may not be the right solution.

I do SingleSSID at home, and also have deployed SingleChannel in both bands - works great in my environment.

 

[L&LD]

Thanks sfx2000.

In the large Enterprise/Carrier space nobody is using Asus routers either.

I offer customers the cheapest cost hardware that they can almost figure out themselves (after some help setting up) that last as long as possible and offer great stability, performance and features. Asus plus RMerlin fits that bill exactly.

In the enterprise space there are contracts required to just use equipment you have already paid dearly for. And nobody I know wants to go down that sinkhole even if the reliability lasted forever. Because even if reliability was 100% a given, the tech keeps marching and what was once great is the bottom of the pile today.

Money per year per device for a given service level is a real issue for customers. As I explain to them and they fully agree with; if an Asus router dies after warranty, simply buy the same or a newer / better one at that time. For some of my previous customers, even if a router died every six months it would be cheaper to replace that device than to have paid for an enterprise solution back then. Now, after almost a couple of years of no need to replace a dead router at all, they are ahead of the game. And if / when their units start failing, replacing devices across their entire network at that time will not seem like a hardship at all, but rather an upgrade that when complete, will still be cheaper and probably give them better performance than the enterprise level hardware and service contracts from so many years ago.

 

[htismaqe]

The thing is, most of us here ARE using Asus or something similar with single SSIDs and it works great.

It seems like you and a couple of other people have dramatically different experiences. In fact, earlier in this thread somebody mentioned that they have lots of problems with single SSID and their iPhone. I have 4 iPhones, 2 iPods, and 1 iPad in this house and all of them "roam" quite well.

 

[htismaqe]

As for service levels, there's definitely a tipping point. I have customers who pay several thousand dollars a year for 24x7x4hr support because 8 hours of downtime has a cost in the MILLIONS of dollars.

 

[L&LD]

I wish I had several of those kinds of customers too.

And they paid all paid a steady percentage of that support cost directly to me each year.

As for service levels, there's definitely a tipping point. I have customers who pay several thousand dollars a year for 24x7x4hr support because 8 hours of downtime has a cost in the MILLIONS of dollars.

 

[azazel1024]

This is like watching people argue how many angels can dance on the head of a pin...

42. Serious, you had to ask?

 

[azazel1024]

I wish I had several of those kinds of customers too.

And they paid all paid a steady percentage of that support cost directly to me each year.

My customers either get my service free as part of my general friendship service plan. Or else I am "paying them", also known as my wife and kids.

Though, I guess since I do help out my parents a little (very little, my Dad is pretty tech proficient), I could be considered to be paying back the large advance they forwarded me.

 

[sfx2000]

42. Serious, you had to ask?

And any good consultant never forgets their towel

Been doing some thinking about devices camping - and this goes back to location of the AP's - outside of enterprise class AP with a WLC, two AP's close enough may not trigger events to handover from one AP to another -

Between RSSI triggers ( <70dB) and deltas between AP1 and AP2 (8 to 12 dB), they might not decide to cutover, or if they do, they might jump to 2.4GHz - and then the same situation happens their on the handup to 5GHz... and CINR (SNR) might come into play there depending on activity

Interesting..

 

[azazel1024]

I've been playing with my APs a little bit and I can trigger bad behavior. Both situations you've described. If they are relatively close together and you can locate yourself far from the APs, they don't do much switching to the stronger one and always on 2.4GHz. Example, if the APs are located on the same floor a room or two away, if I go outside, both have medium to low signal strength and you don't really roam between them unless you go clear to the side of the house. One is clearly stronger than the other one, but it is only 10dB or...but since the signal strength is down in the high 60's and low 70's, 8-10dB would be a nice gain, but my clients won't roam. Placed where I place them across the house from each other, roaming is well behaved.

Can also do the same thing inside when they are a couple of rooms apart on the same floor, devices will hang on to the existing AP even when sitting next to the other one. The issue there is that the RSSI is still high on the connected one, mid 50's, but the one right next to the client is around 40.

So proper placement of the APs are key.

 

[coxhaus]

I would think if you have too much overlap with wireless APs your roaming is not going to be good. You need to turn down the power to create a good roaming environment. The people which have issues here are not tweaking their wireless system enough.

 

[sfx2000]

Can also do the same thing inside when they are a couple of rooms apart on the same floor, devices will hang on to the existing AP even when sitting next to the other one. The issue there is that the RSSI is still high on the connected one, mid 50's, but the one right next to the client is around 40.

So proper placement of the APs are key.

Yep - again, the rule of them I've used in the past is that handhelds are about 30Mw (15 dBm) with zero antenna gain in 2.4GHz and 5GHz - prop loss is about -6dB higher on 5Ghz vs. 2.4Ghz at a given distance and devices tend to start looking at the link around -70 dB otherwise they will camp as they really have no reason to move if they don't have to.

(Laptops, at least most of them, are around the same - 15dBm, but generally more gain on the antenna taking them to around 17-19dBm)

One thing to look for in the wirecaps - 802.11 Probe Requests - this means that the client is searching for potential AP's that it might want to hand off to...

So if it sees one at -70 dB, and another at -62 (or better) it'll likely try to jump... and this can also happen on handup from 2.4GHz to 5GHz - the client may just decide it's not worth doing (keeping in mind that most mobile phones are using BT/WiFi on the same Baseband/RF, antenna's are optimized there for 2.4GHz, and if BT is in use, the MAC/Baseband may decide not to move to 5GHz at all).

Not a certainty, but something to consider with placement of AP's in a roaming network.

And higher Tx power on the AP, or high gain antennas can actually cause more problems, as we really don't want to be about about 21 dBm on the AP - sometimes less is actually more...

 

[azazel1024]

One thing you can do with higher antenna gain, is reduce the radio power on the router. So long as your background interference is low, this can be a big bone from client Tx side performance. Most clients are fairly unbalanced, in part because their radios tend to be 3-8dB weaker than the basestation they are connected to. A few extra dB of gain with the router radio power turned down a few dB can help balance this better while retaining good roaming characteristics.

I tend to notice roaming occurring around -65dBm with iOS handsets and tablets and most windows clients with the drivers set to most aggressive it is -60dBm and in most cases you need at least +10 and sometimes +15dB on the other basestation before there is a transition.

Rules of thumb can always be broken however.

 

[sfx2000]

One thing you can do with higher antenna gain, is reduce the radio power on the router.

I have to respectfully disagree - not more gain on the AP, or more power (talking to the general audience). It's about what the client sees period... not the AP - one must take a systems approach

The rule of thumb in roaming networks is 2x Client Tx/Rx - so dealing with clients there - 15 dBm (30 mW) EIRP suggests that 60 dBm is ideal, no more that 100 mW at the end of the day in a roaming network...

Less is more... and in a roaming network, one must take a client perspective as a priority...

 

[sfx2000]

Rules of thumb can always be broken however

My thumbs have been broken from time to time, lol...