access points downing wired network?

If anyone can offer me any suggestions for a problem I'm having with a
wireless network, I'd sure appreciate it.

The situation is this: a local hotel has had it mandated that they
need to offer wireless access to their high-speed internet connection.
The general manager bought a wireless access point and had us come in
to help him configure it. We got it set up; however, because of the
construction of the hotel its coverage area was not great. The GM
ordered several more access points and we set them up evenly spread
through the length and height of the hotel. There are 8 in total.
They are connected via Cat-5e cabling to his wiring closet, to a
Linksys 24-port switch. Also connected to this switch are their HPNA
switches (which provide network access through their existing phone
wiring) and their router, which I believe is an Efficient DSL router.

Now, whenever someone uses the 802.11b network, after a time all
in-use ports on the switch begin to flicker rapidly and in unison.
When this happens the network is unusable. To correct the situation,
one of two things needs to happen: either the switch needs to be
power-cycled or anywhere from three to seven of the access points need
to be disconnected from the switch (different ones each time, it
seems). I'm not sure if this indicates collisions or just some kind
of chatter or garbage packets that shoot the switch's utilization up
so high that meaningful packets are unable to get through.

As far as how the wireless network is set up, I've tried several
approaches. We're using 128-bit WEP encryption. I've tried using the
same SSID and channel on all the access points and I've also tried
using different channels and SSIDs. The wireless access points
themselves are located in hallways above drop ceilings. We've tried
to arrange the access points and point the antennae in such a way as
to minimize potential interference.

Wireless networking is somewhat new to me, and most of the information
I've been able to find online is either elementary stuff that hasn't
really been helpful or so technical it makes my brain hurt. If anyone
can suggest any resources to me or perhaps share any similar
experiences and their resolutions I'd appreciate it.

Thanks in advance,

Jim Behymer
Service Manager
Computer Works

Hi Jim,

There could be a couple things happening here, both of which I've
experienced in real life situations and are the things to look at.

The first possible scenario is that the access points you're using don't
have a lot of intelligence to them. Lets say there are 3 access points,
A, B, and C, spread through a building, all running on the same channel.
Lets further say that all 3 are poorly designed in the firmware, so that
they don't implement any sort of sanity checking. If they're functioning
as ethernet bridges, what can happen is that access point A will get a
packet on the LAN and broadcast it. B & C will pick it up over the air
and put it back onto the LAN, where A may then pick it up again. It
doesn't take a lot of traffic, then, to destroy everything.

Scenario two, that I have also seen, involved ARP packets. What would
happen is that a machine connected on the same segment as A will do an ARP
request, some of the same magic as above will happen, and the network will
get flooded with ARP requests/replies before the first useful packet is
ever sent.

Both of problems require that all of the APs be connected to the same
physical LAN. In switches, the same thing can occur, which is the point
behind Spanning Tree Protocol. If you can play with it and figure out
that one of these (most likely scen. 1) is the culprit, you can switch the
APs to use different channels. In infrastructure mode, the WICs should
choose the strongest signal and just use that channel.

If my assumptions are incorrect, please let me know and I'll think a bit
harder on it : )

Wes



On Thu, 03 Jul 2003 09:58:39 -0500, Jim Behymer wrote:


> If anyone can offer me any suggestions for a problem I'm having with a
> wireless network, I'd sure appreciate it.
>
.....
>
> Thanks in advance,
>
> Jim Behymer
> Service Manager
> Computer Works

On Thu, 03 Jul 2003 22:48:12 -0400, Yhetti <(E-Mail Removed)> wrote:

>Hi Jim,
SNIP
>
>Both of problems require that all of the APs be connected to the same
>physical LAN. In switches, the same thing can occur, which is the point
>behind Spanning Tree Protocol. If you can play with it and figure out
>that one of these (most likely scen. 1) is the culprit, you can switch the
>APs to use different channels. In infrastructure mode, the WICs should
>choose the strongest signal and just use that channel.
>
>If my assumptions are incorrect, please let me know and I'll think a bit
>harder on it : )
>
>Wes
>
Can co located AP support some spanning tree protocol, to stop
broadcast storms???
>
>
>On Thu, 03 Jul 2003 09:58:39 -0500, Jim Behymer wrote:
>
>
>> If anyone can offer me any suggestions for a problem I'm having with a
>> wireless network, I'd sure appreciate it.
>>
>....
>>
>> Thanks in advance,
>>
>> Jim Behymer
>> Service Manager
>> Computer Works

These sound like good suggestions. Here's something else to
consider--could you simplify this whole thing by using a single WAP,
adding a 1-watt bidirectional amp to it, and choosing a directive
antenna array located out in the courtyard? There are sectoral arrays
that are configured to deliver a broad horizontal pattern while having
essentially no radiation to the rear. I realize that an amp costs
about as much as 3 WAPs, but it might be a simple solution.


On Thu, 03 Jul 2003 22:48:12 -0400, Yhetti <(E-Mail Removed)> wrote:

>Hi Jim,
>
>There could be a couple things happening here, both of which I've
>experienced in real life situations and are the things to look at.
>
>The first possible scenario is that the access points you're using don't
>have a lot of intelligence to them. Lets say there are 3 access points,
>A, B, and C, spread through a building, all running on the same channel.
>Lets further say that all 3 are poorly designed in the firmware, so that
>they don't implement any sort of sanity checking. If they're functioning
>as ethernet bridges, what can happen is that access point A will get a
>packet on the LAN and broadcast it. B & C will pick it up over the air
>and put it back onto the LAN, where A may then pick it up again. It
>doesn't take a lot of traffic, then, to destroy everything.
>
>Scenario two, that I have also seen, involved ARP packets. What would
>happen is that a machine connected on the same segment as A will do an ARP
>request, some of the same magic as above will happen, and the network will
>get flooded with ARP requests/replies before the first useful packet is
>ever sent.
>
>Both of problems require that all of the APs be connected to the same
>physical LAN. In switches, the same thing can occur, which is the point
>behind Spanning Tree Protocol. If you can play with it and figure out
>that one of these (most likely scen. 1) is the culprit, you can switch the
>APs to use different channels. In infrastructure mode, the WICs should
>choose the strongest signal and just use that channel.
>
>If my assumptions are incorrect, please let me know and I'll think a bit
>harder on it : )
>
>Wes
>
>
>
>On Thu, 03 Jul 2003 09:58:39 -0500, Jim Behymer wrote:
>
>
>> If anyone can offer me any suggestions for a problem I'm having with a
>> wireless network, I'd sure appreciate it.
>>
>....
>>
>> Thanks in advance,
>>
>> Jim Behymer
>> Service Manager
>> Computer Works

On Thu, 03 Jul 2003 22:48:12 -0400, Yhetti <(E-Mail Removed)> wrote:

>Hi Jim,
>
>There could be a couple things happening here, both of which I've
>experienced in real life situations and are the things to look at.

<snip>

Thanks for the help, Wes. I'm going to implement some changes based
on your suggestions and on some things I've learned about the 802.11b
channel definitions and hopefully that will clear up the problem.
When I've learned something definitive, I will re-post.

Thanks again,

Jim Behymer
Service Manager
Computer Works