Isolation on Multiple SSIDs

[b1ggjoe]

For those of you with multi SSIDs, are you implementing any sort of isolation?


Sent from my iPhone using Tapatalk

 

[htismaqe]

I only have it enabled on the Guest Network.

 

[sfx2000]

For those of you with multi SSIDs, are you implementing any sort of isolation?


Sent from my iPhone using Tapatalk

Depends on the Use-Case for deployment - AP isolation for a single AP means than clients would have to hair-pin in for internal services if exposed beyond the AP level (for most home AP's that means out to the internet and back).

This may or may not be expected - depends on network design.

Most folks don't need mulitple SSID's - even for Guest Networks... but the option is there..

 

[sfx2000]

I use several SSID here.

Desktops, laptops, netbooks (have their own)
Tablets and smartphones (have their own)
Wii and Network Media Players, Blue Ray with Internet (have their own)

Up to you - every SSID on a common channel/mac takes 12 percent of bandwidth... that means each one consumes that much more at 802.11b data rates...

beacon frames are always at 802.11 minimum data rates...

Best case for common SSID's I can think of...

sfx

 

[roldogg]

I'm running an Asus RT-AC66U, SSID for the 2.4GHZ and a different SSID for the 5GHZ. Two guest networks are enabled (just one 2.4GHZ and one 5GHZ), which use the same channel as the main wireless networks on the router, but they have different SSIDs and AP Isolation since my guest network isn't password protected. I also have a Linksys EA4500 at the other end of the house setup in bridge mode and the 2.4GHZ & the 5GHZ networks are setup to use the same SSIDs and the same channels as the main wireless networks on my RT-AC66U, so everything on those networks can see each other and I don't have to change networks every time I go to the other end of the house. Is this the best way to have this setup performance wise or should the EA4500 be setup on different channels and using different SSIDs that the Asus router?

Sent from my SM-P600 using Tapatalk

 

[stevech]

Up to you - every SSID on a common channel/mac takes 12 percent of bandwidth.
sfx

The beacon payload is tiny - a few microseconds.
I would think that 12 percent of capacity has to be incorrect by an order of magnitude or more.

 

[sfx2000]

The beacon payload is tiny - a few microseconds.
I would think that 12 percent of capacity has to be incorrect by an order of magnitude or more.

If you truly believe that, then you don't know 802.11 at all...

I'm not trying to be mean spirited here - beacon frames are required, they're at 802.11 data rates/modulation schemes - the base rates so that legacy clients can demodulate them... and they're fairly large.

I've got time in IEEE 802.11 as well as 802.16... do you?

 

[stevech]

yes. 15 years. R&D and field work.
Let's not impose on readers' here since the esoterica of WiFi system engineering matters-not to 99% of the forum's users.

The 802.11 beacon is not mandatory but is normally used and is essential for enterprise-type WiFi systems.

In this forum, we talk of users' consumer hardware, not enterprise grade in which we'd find longer beacon durations due to more payload (e.g., adjacent stations lists, etc.) that are in proprietary formats.

In 802.11b the beacon frame duration is longer than in newer versions of the standard where the short frame is used. Even so, the '11b beacons are much less than 10% of the 20MHz channel capacity at the beacon's modulation order. That order in 802.11 is the "least common denominator" for backwards compatibility in the standards. The channel capacity term is of course w.r.t. some single or mix of modulation orders among APs and client stations.

In consumer WiFi, the beacon payload is the bare minimum required by the IEEE standards, and the short preamble is most common today.

Note that the term "bandwidth" is misused by lay persons to mean speed or available speed. The term actually means the channel size in Hz under some defined power density spread such as the FCC's rules based on channel masks in the ISM bands.

Yes, having 10 in-range APs on/near the same channel w/beacons will somewhat negatively impact user device throughput. And adjacent channel interference may exist. But this is worst case of about 350 bytes sent as one beacon frame every 0.1024 seconds.

The now-dead Muni-WiFi (as attempted by Earthlink at large scale, using Tropos' hardware), is where this channel utilization by many outdoor high-spot APs' beacons is truly an issue. Most large scale Muni-WiFi is gone now, due to the business model (predictably) proving non-viable.


end of geek-talk

 

[sfx2000]

The 802.11 beacon is not mandatory but is normally used and is essential for enterprise-type WiFi systems

Beacon frame (don't get this confused with SSID broadcast) is absolutely essential for a single AP WLAN - this is is the overhead needed for the non-adhoc network to function - provides system timing, overhead parameters, capabilities, etc...

Yes, you can have a 802.11 channel without Beacon's - but not in normal situations - most of these are proprietary solutions in 802.11b space - I've got PCAP's I can show you to confirm this.

In any event, it's all good - and yes... I'm correct in my assertion. And each additional SSID generates an additional beacon frame, as one cannot put more than one SSID in a single beacon frame. And each Beacon Frame transmitted takes time, not bandwidth, but leaves much less time for actual data frames between the AP and STA's

Sometimes I get carried away... can't help it - engineer, designer, deployer, and yes, standards geek

Steve - you're one of the smartest guys on this board - and for the most part, we're in complete alignment.

best

sfx

 

[azazel1024]

Looking at a nice handy beacon overhead calculator per AP/per SSID on the same channel the overhead is approximately 3.22% with long preamble, 380b beacon frame size and 102.4ms beacon interval.

You can also change the AP to short preamble, which should halve that overhead as its transmitted at 2Mbps IIRC (instead of 1Mbps of long preambles). It also reduces the overhead on other packets too, as the preamble is transmitted faster. Its small, but it should increase overall network throughput by a percent or two.

Only some 802.11b devices require long preamble, some can do short preamble. All 11g and newer devices support short preamble (and possibly all 11a too? Not sure about that).

In short, unless you need legacy compatibility with really old wireless devices, you can change your AP to short preamble and halve the overhead. You could also increase the beacon invertval to something longer, but that can increase the amount of reassociation time when roaming between access points both because of the longer beacon, but no guarantee that the client will pick up the first beacon that gets sent.

If you really, absolutely must run a ton of SSIDs or APs on the same channel, you can always increase the beacon interval some and I deffinitely recommend short preamble unless you have some darned ancient wireless clients laying around that absolutely must connect (I think I have two 11g clients, my Wii and my wireless printer...the former of which has its wireless disable and the other which is wired in to my network. So in effect, its 11n devices only).

Though...when it comes down to it, unless there is a very good reason, you should really only be running two SSIDs at most, your main one and a guest SSID (if you care to). I only run a single SSID. If a guest needs access to wifi I'll just give them my WLAN password. Its no biggie to me. Everything has access controls anyway, so I don't need to worry about wireless isolation. I am sure it makes it an intsy bit easier for them to get unathorized access to my server...but really not that much easier.

 

[sfx2000]

I've seen revolution wifi's worksheet - it's a good place to start - but it's only looking at one side of the link - you need to consider number of STA's on each SSID, and the additional link maint and RTS/CTS overhead from the STA's when running multiple SSID's - so I stand by my numbers.

Aruba has a great whitepaper on what additional SSID's can offer and the impact (along with performance scaling).

IIRC - 5GHz is always short preamble, long preamble was a 802.11b legacy thing...

sfx