Searching \ for '[EE] Explaining WiFi to the technically challenged' in subject line. ()
Make payments with PayPal - it's fast, free and secure! Help us get a faster server
FAQ page: www.piclist.com/techref/index.htm?key=explaining+wifi
Search entire site for: 'Explaining WiFi to the technically challenged'.

Exact match. Not showing close matches.
PICList Thread
'[EE] Explaining WiFi to the technically challenged'
2005\12\16@140421 by Mike Hord

picon face
I'm wondering if anyone can point me to a website
that'll give me some good tools for explaining WiFi
limitations to those without a fair idea of the
underlying technology.

Full story:
My college has begun equipping incoming students
with TabletPCs.  There is some argument now about
the need for wired network connections at each seat
in the classrooms- obviously, some cost is associated
with providing that, but as massive classroom
renovations are scheduled soon anyway, it's not such a
huge deal.

Some people (including those who would claim to be IT
professionals and should know better) are suggeting that
a wireless network is adequate for our needs- namely,
120 students in one room at one time, accessing network
resources.  They argue against our managing the wireless
infrastructure, as well, saying that having the "experts"
from central campus IT install and maintain "enterprise
level" WAPs in each classroom will make the system
somehow better than our doing it ourselves.

The first real stress test occured yesterday.  In a room
set up by the "experts" with "enterprise level" equipment,
120 students turned on their computers and simultaneously
attempted to load a 6 MB slideshow from a shared drive.
It was pandelerium.  I was on hand to provide assistance,
and spent ten minutes running from one person to the next,
advising patience, until some people's file finished loading
and the network's bandwidth slowly thawed out and let some
data through.

The bottom line, what I can't get people to understand, is
that only so many WAPs can be in one place without
interfering with one another, and that interference brings
the bandwidth available to a particular WAP down.  For
some reason, the general understanding is that these
"enterprise level" WAPs can have multiple devices at the
same frequency without causing a problem.  I've tried
explaining it.  I've drawn diagrams, and explained
signal-to-noise ratios and collisions and bandwidth
sharing, and no one listens.  They just nod politely, and
then at the next meeting, they dig in and say "we need
to get the experts out here".

The classroom in question has four WAPs, which means
overlap out of the gate, but I'm not sure how bad it is.
Best case, that gives us 216 Mbps to spread out among
120 students.  Factor in overhead and the worsened S/N
ratio caused by overlap in the frequency bands, and I
doubt we're doing that well.  Now we have a prof who, for
next semester, wants to use a remote desktop program
to control the content on every screen of all 120 students.
I think he's being over optimistic.  In a big way.

Mike H.

2005\12\16@142202 by Alex Harford

face picon face
On 12/16/05, Mike Hord <spam_OUTmike.hordTakeThisOuTspamgmail.com> wrote:
>
> The first real stress test occured yesterday.  In a room
> set up by the "experts" with "enterprise level" equipment,
> 120 students turned on their computers and simultaneously
> attempted to load a 6 MB slideshow from a shared drive.
> It was pandelerium.

What about a bittorrent client with a user friendly wrapper around it?
I'm not too familiar with 802.11 networking but I believe that
devices can talk to each other directly without going through the AP.

Professors controlling the students displays sounds like a security
nightmare to me!

Alex

2005\12\16@143357 by Danny Sauer

flavicon
face
Alex wrote regarding 'Re: [EE] Explaining WiFi to the technically challenged' on Fri, Dec 16 at 13:26:
> Professors controlling the students displays sounds like a security
> nightmare to me!

If they're just subscribed to a display he controls, ideally using a
client which supports multicasting, that's not a huge deal.  If he's
actually controlling theirs, I give it a half hour before the other
students figure out how to disable their own client + enable
themselves to control other students'.  "Teacher, Johnny's browsing
porn again!"  Heh, that's funny every time I imagine it.

--Danny

2005\12\16@144539 by Harold Hallikainen

face picon face
Interesting situation... Are all the WAPs in a room on different channels?
Seems like that'd spread the load out. Having everyone load a 6MB file
simultaneously seems like quite a load. Could the slide show be separated
out do separate files? Wireless is nice in its simplicity of
"installation" (no searching for ethernet cables, etc.), but a wired
system could handle more traffic assuming there's adequate upstream
bandwidth above the router. Maybe a mix of the two (wired for everyone who
has an ethernet connector on their computer, wireless for the rest)? Or,
how about real low power WAPs, forming a cellular network. I don't think
the WLAN cards in the laptops would back off their power, though, so even
though the WAPs would not interfere with each other, the WLAN cards in the
machines probably would. Lotsa fun!

Harold


--
FCC Rules Updated Daily at http://www.hallikainen.com

2005\12\16@152135 by Mike Hord

picon face
I'll address all three responses I've seen...

> Interesting situation... Are all the WAPs in a room on different channels?
> Seems like that'd spread the load out. Having everyone load a 6MB file
> simultaneously seems like quite a load. Could the slide show be separated
> out do separate files? Wireless is nice in its simplicity of
> "installation" (no searching for ethernet cables, etc.), but a wired
> system could handle more traffic assuming there's adequate upstream
> bandwidth above the router. Maybe a mix of the two (wired for everyone who
> has an ethernet connector on their computer, wireless for the rest)? Or,
> how about real low power WAPs, forming a cellular network. I don't think
> the WLAN cards in the laptops would back off their power, though, so even
> though the WAPs would not interfere with each other, the WLAN cards in the
> machines probably would. Lotsa fun!

I haven't really seen any portable PCs in quite awhile without LAN
ports.  Or WiFi, for that matter.  And, since (in the future, at least)
we will control the model of PC to be used, we can pretty well
guarantee the presence of a LAN port.

My initial investigation is suggesting that of the four WAPs, two have
the same channel and the other two are on two other channels.  That
is, there are two channel 1 WAPs, one channel 6 WAP, and one
channel 11 WAP.  I'm not ENTIRELY certain of how two WAPs on
the same channel in close proximity will behave, but my understanding
is that the best one can hope for is that they'll figure it out and each
one gets some portion of a time slot affair.  They can't both talk at
once, because they're on the same channel.  And if they can't both
talk at once, it seems to me that having two going on the same
channel in the same space is WORSE than having one, because of
the overhead associated with tracking whose turn it is to talk, and
routing traffic, and all the other crap that is associated with multiple
devices on one channel.

We do have the possibility of gigabit ethernet from these classrooms
back to the servers.  I think if we put 120 students on switches back
to gigabit, we'd be a far sight better than we are right now.

>From Danny:

> If they're just subscribed to a display he controls, ideally using a
> client which supports multicasting, that's not a huge deal.

Not sure how it works.  I'll investigate more soon.  It does support
the prof being able to directly control one PC, or look at what they
are doing, or force his display onto theirs.  Haven't got the full
details yet.

>From Alex:

> What about a bittorrent client with a user friendly wrapper around it?
> I'm not too familiar with 802.11 networking but I believe that
> devices can talk to each other directly without going through the AP.

I was going to thumbs down this idea, but now that I think about it,
it doesn't sound that bad as a test distribution method.  After all, the
test would then not be available until the prof enters the room, powers
up her PC, and makes the torrent available.  Yes, 802.11 client devices
can form "ad hoc", peer-to-peer networks, but that requires some
changes to the current network settings, which makes it unlikely to
work in such a situation.

Thanks, guys!

Mike H.

2005\12\16@162426 by M Graff

flavicon
face
Mike Hord wrote:
> The first real stress test occured yesterday.  In a room
> set up by the "experts" with "enterprise level" equipment,
> 120 students turned on their computers and simultaneously
> attempted to load a 6 MB slideshow from a shared drive.
> It was pandelerium.  I was on hand to provide assistance,
> and spent ten minutes running from one person to the next,
> advising patience, until some people's file finished loading
> and the network's bandwidth slowly thawed out and let some
> data through.

With 120 students downloading 6 MB files, that's 720,000,000 megabytes.
 Since at most you can expect 50% efficiency in a wireless network due
to the fact that packets are ack'd and so on, you're probably hitting a
max of about 600 kbytes per channel, leaving a very long download time
when there is massive contention.

Even 10 Mbps wired networks have issues with this -- I once had to teach
instructors not to say "OK, now everyone type in your log in name, but
don't press enter.  OK, now press enter.  Type your password, but don't
press enter.  OK, now everyone press enter." and the network would lock
up tight for a good 15 minutes while 45 DECstations would all talk over
NFS to start up their window managers.

> The bottom line, what I can't get people to understand, is
> that only so many WAPs can be in one place without
> interfering with one another, and that interference brings
> the bandwidth available to a particular WAP down.

A channel bleeds over onto other channels.  If I put my wireless card
into "monitor" mode on channel 6, I can see some wireless beacons on
channels 5 and 7, and at times, 4 and 8.

The only US channels that are totally non-overlapping are 1, 6, and 11.
 This means that at best you can have three non-overlapping APs.  Some
APs can share the channels -- the Proxim AP-2000's can.  I understand
this to mean, from their explanation, that each AP coordinates so the
frequency hopping is done differently, so there is little overlap.  How
they manage this in a 802.11g environment, or in a mixed vendor
environment, I have no idea.  I've also seen no traces of these fabled
coordination packets.

> Now we have a prof who, for
> next semester, wants to use a remote desktop program
> to control the content on every screen of all 120 students.
> I think he's being over optimistic.  In a big way.

In short, this cries out for multicast.  Not unicast.

With unicast, you transmit 6 MB * 120 seats.  With multicast, you
transmit 6 MB * 1, with some retransmission due to lost or dropped
packets.  Note that many APs will use a much slower transmit speed for
multicast or broadcast packets, so you may need to crank this up from
the minimum of 1 megabit/sec to 5.5 or even 11 megabit/sec.

--Michael

2005\12\16@163602 by John Ferrell

face picon face
I spent a couple of years in the early 70's trying to deliver what the IBM
salesman sold to Ohio State University. Whatever the magnitude of the
increased compute power, the student population could overwhelm it in two
weeks. Smaller advances were not even observed. Only system stats would show
any gains. I don't think it has changed much.

Think about it. Does anyone really NEED 6 MB of data in a hurry?
If they do, it better be in place before the class starts...

Resource rationing is essential when you are dealing with greedy consumers.

A University environment is the closest thing I ever saw to an infinite
load.

The dialog starts out " you have x megabytes of data transfer per second
under ideal conditions, spend it wisely....."

John Ferrell
http://DixieNC.US

----- Original Message -----
From: "Mike Hord" <.....mike.hordKILLspamspam@spam@gmail.com>
To: "Microcontroller discussion list - Public." <piclistspamKILLspammit.edu>
Sent: Friday, December 16, 2005 2:04 PM
Subject: [EE] Explaining WiFi to the technically challenged



2005\12\16@173721 by Tim ODriscoll

flavicon
face
On Fri, 2005-12-16 at 13:04 -0600, Mike Hord wrote:
> Some people (including those who would claim to be IT
> professionals and should know better) are suggeting that
> a wireless network is adequate for our needs- namely,
> 120 students in one room at one time, accessing network
> resources.

This probably isn't much help right now, but; if the classrooms are
close to eachother, you might get some interference if more than one
goes wireless.. If there's renovations going on soon, it could be an
idea to get some wifi shielding in place....

Cheers,

Tim.

2005\12\16@173944 by William Chops Westfield

face picon face
On Dec 16, 2005, at 11:04 AM, Mike Hord wrote:

> The bottom line, what I can't get people to understand, is
> that only so many WAPs can be in one place without
> interfering with one another, and that interference brings
> the bandwidth available to a particular WAP down.  For
> some reason, the general understanding is that these
> "enterprise level" WAPs can have multiple devices at the
> same frequency without causing a problem.

Hmm.  My understanding matches the "general understanding", and in
fact we have some corporate classrooms capable of handling 100+
people set up with wireless and I don't THINK that there have been
major problems with it, although I don't know that it's had a stress
test like 120 people downloading the same 6MB file at the same time
(more like people reading their email to keep from falling asleep...)
(once upon a time, one could expect a 10Mbit ethernet not conceptually
different from an 802.11 wireless network to handle 120 users without
much problem, but I suppose those days are over.)
But I'm not an wireless expert.  I'll ask around and see if we've
got anything relevant to say...

Are you sure it's the wireless net that is causing problems?  120 users
simultaneously accessing one network disk sounds like a bit of a recipe
for disaster regardless of networking technology...

BillW

2005\12\16@175307 by Marc Lavallée

flavicon
face
Le 16 Décembre 2005 16:37, John Ferrell a écrit :
> Think about it. Does anyone really NEED 6 MB of data in a hurry?

Who needs more than 512K of RAM? ;)

> A University environment is the closest thing I ever saw to an infinite
> load.

Undergrads (and most of their "not so wise" teachers), were raised watching StarTrek, where energy and access to resources is not an issue. Ecology?  The Kyoto agreement? That's for wussy tree huggers; our good kids are not like those ugly ecofreaks, and they will save the human race by going into space! Until then, they are just practicing their wormhole travelling skills by downloading teragigs of hd video on their cell phones. Life will be good no matter what.

> The dialog starts out " you have x megabytes of data transfer per second
> under ideal conditions, spend it wisely....."

Or "you have one gallon of gaz per week".
Then watch the suicide rate going up...
--
Marc

2005\12\16@190334 by William Chops Westfield

face picon face
On Dec 16, 2005, at 11:04 AM, Mike Hord wrote:

> what I can't get people to understand, is
> that only so many WAPs can be in one place without
> interfering with one another...

I don't know if any of the following is applicable to non-cisco
gear, but here's some advice...

{Quote hidden}

The idea of turning the transmit power DOWN seems like new and possibly
key data, but I think numbers computer by others make clear that the
wireless network isn't your only problem...

BillW

2005\12\16@195930 by olin piclist

face picon face
Tim ODriscoll wrote:
> If there's renovations going on soon, it could be an
> idea to get some wifi shielding in place....

Or string a lot of CAT5 cable.

******************************************************************
Embed Inc, Littleton Massachusetts, (978) 742-9014.  #1 PIC
consultant in 2004 program year.  http://www.embedinc.com/products

2005\12\16@231935 by Nate Duehr

face
flavicon
face
Mike Hord wrote:
> I'm wondering if anyone can point me to a website
> that'll give me some good tools for explaining WiFi
> limitations to those without a fair idea of the
> underlying technology.

Two major things:

Half-duplex.

802.11b is 11Mb/s and only channels 1, 6 and 11 don't interfere with
each other.

802.11g is 54Mb/s ONLY if ALL users are running 802.11g - if 802.11b
compatibility is required, the whole network drops to b speeds during b
client access.  802.11g also uses the whole band and one AP will always
interfere with another if they're in range of one-another.

Good luck designing a network that will support 160 users with that.

Nate

2005\12\16@232839 by Nate Duehr

face
flavicon
face
M Graff wrote:

> A channel bleeds over onto other channels.  If I put my wireless card
> into "monitor" mode on channel 6, I can see some wireless beacons on
> channels 5 and 7, and at times, 4 and 8.

It's not bleed-over, the spec is designed that way.

> With unicast, you transmit 6 MB * 120 seats.  With multicast, you
> transmit 6 MB * 1, with some retransmission due to lost or dropped
> packets.  Note that many APs will use a much slower transmit speed for
> multicast or broadcast packets, so you may need to crank this up from
> the minimum of 1 megabit/sec to 5.5 or even 11 megabit/sec.

Full-motion real-time video (anything faster than 24fps) is almost
impossible for the number of users he's talking about in an 802.11x
environment.

Even with the best CODEC's in the H.323 standard, (and those aren't
cheap - you don't get them with free tools like Microsoft NetMeeting)
you'd be pushing so much bandwidth to each user station (unless you used
multicast) on a half-duplex network that your contention/collision rate
would skyrocket.

Document conferences might work but their bursty nature would cause
massive pile-ups at either client startup (cached content - beginning of
class period) or at each frame change.

Nate

2005\12\17@151545 by Peter

picon face

Simple answer:

You need 120 people * 5Mbps speed for reasonable operation. That's
'only' 600 Mbps guaraneed continuous bandwidth (not burst) and
120-client simultaneous concurrent access (i.e. at least 120 sessions in
parallel) on any server or AP. At this speed the 6MB presentation would
load in ~12 seconds and your prof would be able to control everyone's
screens in broadcast mode in real time with no trouble. 5MBps is roughly
equivalent with 10x faster than 512kbyte/sec DSL speed.

The next small point: Ethernet and WiFi does not scale linearly with
users and/or speed. Very often twice the number of users means one
fourth of speed (collisions etc).

I think that your listeners can understand these simple numbers.

Peter

2005\12\17@153852 by Peter

picon face


On Fri, 16 Dec 2005, Olin Lathrop wrote:

> Tim ODriscoll wrote:
>> If there's renovations going on soon, it could be an
>> idea to get some wifi shielding in place....
>
> Or string a lot of CAT5 cable.

And a 120-way switch (can't use a router at that ratio) ? And a couple
of feet of gigabit ethernet to connect it to a suitable server in its
own 19" rack ? With a connection to an air conditioner to prevent it
from melting, a UPS, and serious soundproofing so people can hear what
the teacher is saying over the whine of the 12-20 cooling fans in the
rack ?

Peter

2005\12\17@160529 by olin piclist

face picon face
Peter wrote:
> And a 120-way switch (can't use a router at that ratio) ? And a couple
> of feet of gigabit ethernet to connect it to a suitable server in its
> own 19" rack ? With a connection to an air conditioner to prevent it
> from melting, a UPS, and serious soundproofing so people can hear what
> the teacher is saying over the whine of the 12-20 cooling fans in the
> rack ?

I think you're sensationalizing this a bit.  You're not trying to give
everyone a pristine 10Mb/sec network connection, but something more reliable
than WiFi.  With 120 users the total bandwidth requirements are going to
exceed anything you can supply for a reasonable price.  And of course you
don't do this in a single 120:1 step.  8:1 or 16:1 switches are cheap,
available, and don't require much power.  Two tiers of switches should do
this fine.

I would look into throttling the bandwidth of each individual connection
coming out of the second tier switches.  However, I don't know anything
about this and whether such things are cheaply and readily available.  If
each individual line was limited to 1Mb/sec, things should work out well
enough.  I have no idea if there are switches available that do something
like that already or not.  If not, this sounds like a product idea for cases
just like this.  Maybe this can be combined with a small HTTP proxy server
since the web references for a class are likely to be very correlated.  I
don't know how to easily configure each student's computer to that proxy
server just while in that class though.  Oh well, just a thought, and
probably not a good one.

If each student in the class had 1Mb/sec available, that would still allow
most people to get the full bandwidth most of the time.  It's a lot better
than most DSL connections, and should be more than sufficient for anything
you should be legitimately doing in a class.


******************************************************************
Embed Inc, Littleton Massachusetts, (978) 742-9014.  #1 PIC
consultant in 2004 program year.  http://www.embedinc.com/products

2005\12\17@184442 by Mike Hord

picon face
> I think you're sensationalizing this a bit.  You're not trying to give
> everyone a pristine 10Mb/sec network connection, but something more reliable
> than WiFi.  With 120 users the total bandwidth requirements are going to
> exceed anything you can supply for a reasonable price.  And of course you
> don't do this in a single 120:1 step.  8:1 or 16:1 switches are cheap,
> available, and don't require much power.  Two tiers of switches should do
> this fine.

What I'm envisioning would be one or two gigabit lines into the room,
with one or two switches of 16-32 ports each, and one line from each
of those switches to a single 8 port switch handling the traffic from
each row of seats.

Of course, one problem with this idea (which I can't overcome) is
the University telecom policy disallowing passing network cables
from one room into another.  Last time I heard, they considered
cables passed down through one hole in the floor and then back
up, with both ends in the same room, to be an "illegal" setup.
They'll charge us per jack, in that case.

> I would look into throttling the bandwidth of each individual connection
> coming out of the second tier switches.

A good idea, and maybe we could do that with the WAPs.  IF we had
control over their settings.

> If each student in the class had 1Mb/sec available, that would still allow
> most people to get the full bandwidth most of the time.  It's a lot better
> than most DSL connections, and should be more than sufficient for anything
> you should be legitimately doing in a class.

If it was constant, sure.  My cable internet connection is rated at ~1.5
megabits, best case, and it's usually more than fast enough for anything
I'd want to do.  The problem is not just resource limitation, but also user
education.  ANYTHING which depends purely on user education is
(IMHO) doomed to failure.  No matter what we say to the instructors,
sooner or later, one of them will build full motion video into their syllabus,
and won't come to us with the problem until class.  No matter what we
say to the students, when their slideshow doesn't open instantly, they'll
double click it again, and again, until something happens.

Mike H.

2005\12\17@184800 by William Chops Westfield

face picon face
On Dec 17, 2005, at 12:15 PM, Peter wrote:

> You need 120 people * 5Mbps speed for reasonable operation.

I would say "You NEED multicast."  Alas, while multicast was a hot topic
not so long ago, it sorta got ignored as assorted hardware scaled faster
than expected (but didn't support multicast very well), usage patterns
changed, bubbles burst, and research funding got cut.

> Ethernet and WiFi does not scale linearly with users and/or speed.

A controversial statement.  There were arguments about ethernet vs
token ring revolving around this topic for a long time.  They mostly
went away when 100Mbit ethernet turned out to be orders of magnitude
cheaper than 100Mbit token rings (or even 16Mbit token ring, for that
matter.)

Sigh.
BillW

2005\12\17@190305 by olin piclist

face picon face
Mike Hord wrote:
>> I would look into throttling the bandwidth of each individual
>> connection coming out of the second tier switches.
>
> A good idea, and maybe we could do that with the WAPs.  IF we had
> control over their settings.

But it's too late in the WAPs.  All the higher speed tries and retries and
collisions are already occurring.  Throtteling there will probably just make
things worse.

I was talking about wired ethernet only.  Wireless is clearly silly in this
application.

> If it was constant, sure.

If 100Mb is used between the two tiers of switches, and the tier 1 switch
goes to a gigabit backbone, then everyone can have their 1Mb/sec
simultaneously.

> The problem is not just resource limitation, but
> also user education.  ANYTHING which depends purely on user education
> is (IMHO) doomed to failure.

That's why you throttle each of the 120 taps individually.  No matter what
the professor asks them to do or each student does, an individual student's
load can't exceed 1Mb/sec (which is faster than the DSL link I'm using right
now, and which is quite fine for most things I try to do).

> No matter what we say to the
> instructors,
> sooner or later, one of them will build full motion video into their
> syllabus, and won't come to us with the problem until class.

If they don't listen, they'll find out eventually for themselves that it
doesn't work.  At least they wont be creating an enourmous mess for everyone
else.  If the prof wants to show streaming video, this should be done with
the room projector on the big screen in the front of the class.  Sometimes
people seem to forget that a lot of education went on successfully before
there was an internet.  I'm not suggesting going back to the dark ages, but
just because you can do something doesn't make it a good idea.


******************************************************************
Embed Inc, Littleton Massachusetts, (978) 742-9014.  #1 PIC
consultant in 2004 program year.  http://www.embedinc.com/products

2005\12\17@193334 by William Chops Westfield

face picon face
On Dec 17, 2005, at 1:05 PM, Olin Lathrop wrote:

> With 120 users the total bandwidth requirements are going to
> exceed anything you can supply for a reasonable price.

What you really want for this particular situation (if you can't have
multicast :-) is switches with a small amount of caching capability.
If each of the "proximate" 16 port switches had, say, 10MB of cache
capability, you'd be in pretty good shape, even if they were only
100Mbps switches.

Unfortunately, switches come in about two varieties:
  1) Cheap, implemented mostly in HW, and having very little actual
     memory or processing capability...
  2) feature-full and expensive.

Serving up even 16 * 6MB is no trivial task...

BillW

2005\12\17@195549 by William Chops Westfield

face picon face
On Dec 17, 2005, at 3:44 PM, Mike Hord wrote:

> Of course, one problem with this idea (which I can't overcome) is
> the University telecom policy ...
> They'll charge us per jack, in that case.
>
Maybe you should just 'demand' a solution from the powers that be and
see what they come up with.

It would be SO much better to get wireless working than to come up with
a plug-based solution.  RJ45s are just NOT robust enough to have
STUDENTS
plugging in and out several times a day.  You'll have maintenance
nightmares
with just the physical wiring...

perhaps what you really need is a smarter server.  It seems to me that
120 students downloading 6MB ought to take no longer than 15 minutes if
you only let one of them download it at a time, and that even a modest
server and network infrastructure ought to be able to handle several
such downloads at one time.  The 120 simultaneous access are causing
thrashing and pessimal behavior for everyone.  Servers ought not offer
to do more than they are capable of.  (that's SUPPOSED to be built into
assorted protocols like TCP and even "ethernet.")

BillW

2005\12\17@222804 by Jake Anderson

flavicon
face
big post with a quick note for olin at the top
The proxy you want is a "transparent proxy", it just intercepts anything
outbound with no config on the client PC and you can't sidestep it either.

Main Post
You want to be running a samba share (windows file sharing),FTP,HTTP,DHCP,
transparent proxy (eg squid), and access control/content filtering say dans
guardian or URLfilter so they don't sit there leaching porn or swapping
movies (from outside, you cant (easily) stop it from running inside the
network)

The students can copy the files by either FTP (preferred), HTTP (still ok),
samba (not so good but eh they should all be able to do it no matter how
messed up or illiterate they are, also good for applications where a mapped
network drive or a personal information store is handy)

Very quick hardware calculation (big iron style)

5x FS726T netgear 24 port rack mount switches with 2x gbit uplink ports
$375      $1875
2x GS608 8 port Gbit switches (desktop so you may want to go rack)        $
90      $ 180
1x 4Ru Chassis
$217      $ 217
1x 600W PSU
$160      $ 160
1x Tyan Socket 940 Dual CPU M/B K8W (S2885ANRF)
$900      $ 900
1x IBM PCI -X Gbit Copper Ethernet
$130      $ 130
2x Geil Dual Channel PC3200 400MHz DDR, 2.0GB (2x1.0GB)
$352      $ 704
2x AMD Opteron 265 Dual Core, Socket 940, Server
$1194     $2388

$6554
That should be able to provide 2Gbit sustained transfer (roughly).
<maths>
So say 1.8 with overheads your looking at 15mbit per student or 1.85 MByte/s
so 6mb file in 4 seconds or so.
</maths>
It'd also do a fair job serving web pages to them and the like having 4 CPU
cores in total (that also carries a fair geek coolness factor)
Inside a rack with a door you probably wouldn't even hear it outside.
That was the first mbo that came close to the bill, there will be others
with dual GBit on a better buss (that's on a PCI-X so it could become
limited, your better off with one that's in the chipset and plugs straight
into the HT backbone, then you can add another PCI-X or 2). You would also
probably up the quality of the gbit switches to non-blocking, those again
were the first drawn out of the hat.
net cost << 1 students fee for the year ;->

Plan-B small iron expandable.
5x 1RU 3Ghz Xeon 180GB disk and gbit LAN (or 2 I forget exactly now) is
~$1500      $7500
5x FS726T netgear 24 port rack mount switches with 2x gbit uplink ports
$375      $1875

Each switch runs its own network, the 1Ru PC's run a network file system or
something similar so all their data comes from the one source. They each run
their own proxy/firewall/content filter start off with 2x 1RU servers, if
the network runs slowly, add more, lather and repeat.
Little bit noisier, 1Ru computers really work their fans hard.

You only need a generic IDE disks, you are going to be running in ram mostly
so disk I/O should be minimal.
You only need a half rack for these solutions.
Upside you wont need any custom SW on the clients computer, as the clients
are students and most of them don't know their ass from their elbow this is
a bonus ;-> (6th week of digital systems course question asked in lecture
"whets an AND gate")

To do a passable job with wireless you would want 1 AP per say 20 students
with a *very* high gain antenna and low power output and basically set up a
cellular system. Just plonking AP's in wont work (as you have found)
I have seen a "puck" looking AP setup for something like this, it had
something like 10 different AP's in it, all tuned to different specs
(a,b,g,11,54,108 etc) with ceramic directional antennas and the like inside
it, I haven't been able to find the review of it again but it looked like it
might come close to working for you.
nowhere near as good as a wired setup though of course.


Feel free to contact me if you want more information on a setup like this
grooveee (that whole at symbol thing) optushome (place a dot here) com (oh
look another dot) au
> {Original Message removed}

2005\12\17@235256 by Nate Duehr

face
flavicon
face
Olin Lathrop wrote:

> If each student in the class had 1Mb/sec available, that would still allow
> most people to get the full bandwidth most of the time.  It's a lot better
> than most DSL connections, and should be more than sufficient for anything
> you should be legitimately doing in a class.

This is a problem for certain.  The classroom machines shouldn't really
have external Internet access or the instructor will find them more
disruptive than useful, over the long term.

I'm still trying to figure out why an instructor needs a PC in front of
every student to display a screen full of information.

A cheap projector for the slides and a walk around the room to actually
help the students IN PERSON seems like what I'd be paying for if I were
taking a course... not some fancy equipment to display the slides on my
PC and some lazy instructor who can't get off his fat butt to walk back
to my workstation to help me if I raise my hand.

Ask the administrators if this expense is what the students are paying
for, or if it can be done just as easily with a projector on a screen, I
say... (GRIN).

Nate

2005\12\18@044531 by Morgan Olsson

flavicon
face
I think you should really start in the other end.

Do they really need to download that data during the lesson they need it?
That begs for big problems in class, as any lagging on any machine vill make significant delay on the class session.  (Or the teacher have forgot uploading it or any problem with network, or other network/server load or maintenance coincide, or whatever.  Just-in-time often is definitely not htat, especially if you start not until you already needed it...

Why can´t they be given th alink to file to download until NEXT lesson, and in current lesson use what they have downloaded from th elink viven the lesson before.  Then they can download during lesson, or any time else.  Much less hassle not only for IT department.

And of course look at if they really need such large file.

Also I would not like to sit in a room with plenty of RF devices pumping, plus all RFI from computers already.. There are different reports of health effects on that, and until tha tis solved we shuld definitely not force pupils into theede maybe hazardous environment anyway...  (Says I sitting an elactronic lab...)  ;)

/Morgan
--
Morgan Olsson, Kivik, Sweden

2005\12\18@102856 by Howard Winter

face
flavicon
picon face
On Fri, 16 Dec 2005 13:04:21 -0600, Mike Hord wrote:

>...<
> Some people (including those who would claim to be IT
> professionals and should know better) are suggeting that
> a wireless network is adequate for our needs- namely,
> 120 students in one room at one time, accessing network
> resources.  They argue against our managing the wireless
> infrastructure, as well, saying that having the "experts"
> from central campus IT install and maintain "enterprise
> level" WAPs in each classroom will make the system
> somehow better than our doing it ourselves.

And what do these "experts" think is different about an "enterprise" WAP versus an off-the-shelf one?  802.11x
is a standard, and it has limitations, and those don't extend to sending 6MB to 120 users at once!  Ya canna'
change the laws o' physics, captain!  Even by calling it something that has a better marketing sound to it  
:-)  If they are on a single channel, only one WAP or user can be transmitting at any one moment in time, and
if more than one tries, all will fail and will go into a wait-and-retry loop, so reducing the theoretical
throughput dramatically.  They are going back to the days of 10Mb/s hubs on wired Ethernet, rather than the
switches we use nowadays.

> The first real stress test occured yesterday.  In a room
> set up by the "experts" with "enterprise level" equipment,
> 120 students turned on their computers and simultaneously
> attempted to load a 6 MB slideshow from a shared drive.
> It was pandelerium.  I was on hand to provide assistance,
> and spent ten minutes running from one person to the next,
> advising patience, until some people's file finished loading
> and the network's bandwidth slowly thawed out and let some
> data through.

Ah, "Pandelirium" - my "you learn something new every day" item for today :-)  Any of these "experts" should
have known that would happen - you have a half-duplex channel (send or receive at any one time, not both),
which causes a geometric progression of collisions as traffic increases.  It's not rocket science, it's basic
comms which has always been that way with CSMA/CD ever since it was first thought of.  Ask any proponent of
Token Ring (or even ARCnet) and they will tell you why it's a problem.  :-)

> The bottom line, what I can't get people to understand, is
> that only so many WAPs can be in one place without
> interfering with one another

Yes, in the USA it's three, since that is the number of non-overlapping channels (In Europe & Japan it's more
like 4 - I have no idea why the US chopped off the top few channels).  So the best case you could hope for is
3 x 40 users, so in an ideal situation each user would get 1/40th of 54Mb/s, which a rough calculation shows
would mean a bit over a minute and a half to get 6MB to everyone.  But of course the collisions problem makes
this far from ideal, and I think you could reckon on at least 5 minutes in a real situation - maybe much
longer.

> and that interference brings
> the bandwidth available to a particular WAP down.  For
> some reason, the general understanding is that these
> "enterprise level" WAPs can have multiple devices at the
> same frequency without causing a problem.

It may mean they can coexist, but it doesn't mean they can transmit (or receive transmissions for them) at the
same time - just that they will step aside when the other one is transmitting.  When they are in the same
room, it means that you effectively only have one of them because the diversity that would come if they were
at opposite ends of the site is lost - every user machine will receive from both of them.

>  I've tried
> explaining it.  I've drawn diagrams, and explained
> signal-to-noise ratios and collisions and bandwidth
> sharing, and no one listens.  They just nod politely, and
> then at the next meeting, they dig in and say "we need
> to get the experts out here".

If they get real experts, rather than equipment salesmen or bluffers, then they will tell them what you are
trying to say, because that is the real situation.

> The classroom in question has four WAPs, which means
> overlap out of the gate, but I'm not sure how bad it is.
> Best case, that gives us 216 Mbps to spread out among
> 120 students.  Factor in overhead and the worsened S/N
> ratio caused by overlap in the frequency bands, and I
> doubt we're doing that well.

Nowhere near that well - even if there was no interference between WAPs, adding in the overhead for
acknowledging packets would reduce it by at least 10%, and dealing with collisions for an individual WAP would
reduce it much more.

> Now we have a prof who, for
> next semester, wants to use a remote desktop program
> to control the content on every screen of all 120 students.
> I think he's being over optimistic.  In a big way.

Ask him to do the calculations of the bandwidth he needs, and see if he will realise how feasible it is...

Cheers,


Howard Winter
St.Albans, England


2005\12\18@121316 by Nate Duehr

face
flavicon
face
William Chops Westfield wrote:

>> Ethernet and WiFi does not scale linearly with users and/or speed.
>
>
> A controversial statement.  There were arguments about ethernet vs
> token ring revolving around this topic for a long time.  They mostly
> went away when 100Mbit ethernet turned out to be orders of magnitude
> cheaper than 100Mbit token rings (or even 16Mbit token ring, for that
> matter.)
>
> Sigh.
> BillW

Then the users figured out ways to saturate 100Mb/s Ethernet.  ;-)

The discussions will start up again as VoIP and Video applications in
real-time blow up the bandwidth needs, and then Gigabit-speed Ethernet
will be deployed locally and the local problem will go away... but the
interesting question is -- will the carriers and those who put up the
WAN pipes and Internet find it cost-effective to continue building up
the speeds of non-local-area pipes... and how much will they cost the
person ordering them at the user end of the pipes.  :-)

Video real-time applications have been relegated mostly to ISDN here in
the States over circuit-switched networks due to the scarcity of huge
bandwidth pipes, but that will change... over from H.320 to H.323 we go!

Nate, .....nateKILLspamspam.....natetech.com

2005\12\18@122532 by Nate Duehr

face
flavicon
face
Olin Lathrop wrote:

> If they don't listen, they'll find out eventually for themselves that it
> doesn't work.  At least they wont be creating an enourmous mess for everyone
> else.  If the prof wants to show streaming video, this should be done with
> the room projector on the big screen in the front of the class.  Sometimes
> people seem to forget that a lot of education went on successfully before
> there was an internet.  I'm not suggesting going back to the dark ages, but
> just because you can do something doesn't make it a good idea.

"Management" in many larger organizations tend to do whatever they want,
instead of actually delegating the technical questions to the people
they hired to understand the technical issues...

I have run into many situations where the best I could do was to
document why the system wouldn't work, deploy it anyway with
appropriately communicated warnings, and try not to make too many people
mad six months later when you pull out the "I told you so" document and
suggest the correct solution to whatever the problem was, once again.
It can be a tricky tight-rope, politically.

People can be strange sometimes.  Totally ignoring the person you hired
specifcially to handle technical issues is silly, but it happens every
day.  In corporate America, the danger for the technical person is that
the built-in assumption is that the technical person didn't do their job
correctly, and that the manager "knows best", in many organizations.
It's prevalent enough, that it's almost epidemic.  And technical people
that have been burnt by this attitude before will do almost anything,
going out of their way, to stay away from non-technical managers who
don't understand how to properly delegate technical decisions to the
technical professionals.  These non-technical managers can always find a
cadre of similarly non-technical technicians who are clueless but want
to work with the "successful" manager.

I personally think that this is one of the reasons why many good
technical people actively search for work in smaller organizations where
less of this silliness goes on.  Smaller organizations either as
standalone companies, or within larger organizations -- small teams that
are knows as being able to do the hard (and also satisfying) work.

Small organizations tend to have less places where bad managers and or
bad technicians can hide and not be seen as people who really can't
deliver on the tough technical problems with real solutions that are
cost-effective, maintainable, and that make a profit for the company.

Nate, EraseMEnatespam_OUTspamTakeThisOuTnatetech.com

2005\12\18@133157 by Mike Hord

picon face
> And what do these "experts" think is different about an
> "enterprise" WAP versus an off-the-shelf one?  802.11x
> is a standard, and it has limitations, and those don't
> extend to sending 6MB to 120 users at once!  Ya canna'
> change the laws o' physics, captain!

My point, exactly.  We really have two insurmountable
obstacles here:  physics and the US Federal Government.
If the Federales increased the band we are allowed to use
for 802.11, we could add more WAPs.  If physics allowed
us to transmit multiple signals in the same band, we could
add more WAPs.  I leave it to reader to decide which is
more likely.

>  If they are on a single channel, only one WAP or user
> can be transmitting at any one moment in time, and
> if more than one tries, all will fail and will go into a
> wait-and-retry loop, so reducing the theoretical
> throughput dramatically.

AIUI, some WAPs can be setup to negotiate their wait/
retry through the wired network, to reduce the hit from
broadcast collisions.  I don't know how much good that
does- after all, the two WAPs still can't broadcast at the
same time.

> Any of these "experts" should
> have known that would happen

Hence the quotation marks.  I'm not sure how expert their
knowledge really is.

{Quote hidden}

That assumes that users don't inadvertently begin second
or third file transfers by double clicking again, and again,
and again, when the file doesn't instantly open.  It took
about 15 minutes before we got everyone going.

Mike H.

2005\12\18@145852 by Peter

picon face


On Sat, 17 Dec 2005, Peter wrote:

> Simple answer:
>
> You need 120 people * 5Mbps speed for reasonable operation. That's 'only' 600
> Mbps guaraneed continuous bandwidth (not burst) and 120-client simultaneous
> concurrent access (i.e. at least 120 sessions in parallel) on any server or
> AP. At this speed the 6MB presentation would load in ~12 seconds and your
> prof would be able to control everyone's screens in broadcast mode in real
> time with no trouble. 5MBps is roughly equivalent with 10x faster than
> 512kbyte/sec DSL speed.

That was 512 bbit/sec of course.

Peter

2005\12\18@151351 by Peter

picon face

On Sat, 17 Dec 2005, Olin Lathrop wrote:

> Peter wrote:
>> And a 120-way switch (can't use a router at that ratio) ? And a couple
>> of feet of gigabit ethernet to connect it to a suitable server in its
>> own 19" rack ? With a connection to an air conditioner to prevent it
>> from melting, a UPS, and serious soundproofing so people can hear what
>> the teacher is saying over the whine of the 12-20 cooling fans in the
>> rack ?
>
> I think you're sensationalizing this a bit.  You're not trying to give
> everyone a pristine 10Mb/sec network connection, but something more reliable
> than WiFi.  With 120 users the total bandwidth requirements are going to
> exceed anything you can supply for a reasonable price.  And of course you
> don't do this in a single 120:1 step.  8:1 or 16:1 switches are cheap,
> available, and don't require much power.  Two tiers of switches should do
> this fine.

You are right, but I think that a distributed solution would work much
better. F.ex. if there would be 10 inexpensive all-purpose computers in
the classroom, networked, and running mirroring sotware and
samba/http/ftp servers each, then it would all boil down to using one
commercial low cost 16-way router (or USB wireless adapter) per server
to serve 120 people. The local network (server to router to 10-16 users)
would be 100MBps and likely not see contention. The actual material
would be distributed by the mirroring software between the servers. This
should guarantee >5MBps/client no matter what, and due to the tiering
(2-level), it may mean >2.5MBps/client even if the material is injected
'at the last moment' by the teacher's laptop.

Also the cost would be much lower than other solutions, and the 10
servers would be usable for other tasks when not being stress-tested,
they being already networked between them to a single hub. Here by
servers I mean $500 commodity PCs with 2 network cards in them, each.

Serving 16 simultaneous clients using Apache or Samba on Linux or *BSD
is very easy and will be as fast as one can imagine even on a small
machine with 128MB ram and ~1GHz cpu.

Throttling a network is seldomly good. The kind of throtting needed here
is implemented inherently in TCP/IP (backoff & retry etc). Just tuning
the TCP/IP stack parameters of the server may be enough.

There remains to be seen, what a distributed solution would mean for
Wifi (10 APs in a room). Hoping for Aloha and power backoff to work. Has
anyone got experience with 'short range' Wifi ? I.e. deliberately
crippling the range of a Wifi dongle, f.ex. USB, and seeing if a client
would 'switch cells' between two of these in the same room, based on
distance (and standing waves and ...) ?

Peter

2005\12\18@153953 by Peter

picon face


On Sat, 17 Dec 2005, William Chops Westfield wrote:

{Quote hidden}

I think that some stress testing will show that a normal bunch of
machines hitting the same server on the same segment at the same time
will result in spectacularly un-democratic bandwidth distribuition and
huge backoff times which will cause machines that 'lost' the race to
appear frozen from the application pov. . 5 or 6 machines hitting a
server at the same time on a 100MBps segment are enough to demonstrate
this. The 100MBps network can be easily saturated by even the cheapest
machines in use now, i.e. even peer to peer 100MBps through a dumb
router results in less than 100MBps speed and even running two identical
transfers between two machines results in un-democratic bandwidth
distribution (1 server vs 1 client, 2 client programs accessing the
server simultaneously).

Peter

2005\12\18@155230 by Dave Lag

picon face
Nate Duehr wrote:
> The discussions will start up again as VoIP and Video applications in
> real-time blow up the bandwidth needs, and then Gigabit-speed Ethernet
> will be deployed locally and the local problem will go away... but the
> interesting question is -- will the carriers and those who put up the
> WAN pipes and Internet find it cost-effective to continue building up
> the speeds of non-local-area pipes... and how much will they cost the
> person ordering them at the user end of the pipes.  :-)
>
> Video real-time applications have been relegated mostly to ISDN here in
> the States over circuit-switched networks due to the scarcity of huge
> bandwidth pipes, but that will change... over from H.320 to H.323 we go!
>
> Nate, natespamspam_OUTnatetech.com

There are video carriers who operate in the wholesale arena providing
low jitter and latency backbone for video.
The one I know keeps telling me how lousy the cableco "highspeed"
endlinks are.

D


2005\12\18@164028 by William Chops Westfield

face picon face

On Dec 18, 2005, at 12:39 PM, Peter wrote:

>>> Ethernet and WiFi does not scale linearly with users and/or speed.
>>
>> A controversial statement.

> I think that some stress testing will show that a normal bunch of
> machines hitting the same server on the same segment at the same time
> will result in spectacularly un-democratic bandwidth distribuition and
> huge backoff times

But are those ethernet limitations or transport limitations?  There was
a
lot of research on fairness and so on at the TCP level.  Briefly.  Then
microsoft got into networking in a big way with proprietary code that
(probably) ignored much of that research and the rest is history...

BillW

2005\12\18@183941 by Peter

picon face
William "Chops" Westfield <westfw <at> mac.com> writes:

> On Dec 18, 2005, at 12:39 PM, Peter wrote:
>
> >>> Ethernet and WiFi does not scale linearly with users and/or speed.
> >>
> >> A controversial statement.
>
> > I think that some stress testing will show that a normal bunch of
> > machines hitting the same server on the same segment at the same time
> > will result in spectacularly un-democratic bandwidth distribuition and
> > huge backoff times
>
> But are those ethernet limitations or transport limitations?  There was
> a
> lot of research on fairness and so on at the TCP level.  Briefly.  Then
> microsoft got into networking in a big way with proprietary code that
> (probably) ignored much of that research and the rest is history...

My reference to testing refers to unix:unix testing. No uncompliant tcp/ip stack
was involved. My understanding is that assuming 2 machines hit the same server
with a request, and the packets collide, the clients will re-request, subject to
a random delay. This is when one of the clients wins. Thereafter the
first client starts pulling data and saturates the network, and the second
client retransmits at longer and longer intervals, with very good chances of
causing a collision followed by an even longer delay to the next request.

It is possible to control this by using throttling but it's still not so good,
since the peers are not synchronised, the higher the traffic the more
collisions, the more retransmit delays, and the slower it gets. I think that a
segment with a lot of peers can only run at up to 50% of the segment speed, and
this can only be divided evenly if per-client throttling is used (which
throttling must be dynamic - i.e. one peer - 100% bw - 2 peers - give 50% each -
3 - give 33% each etc). If any peers engage in unruly behavior, like ping or arp
storms or broadcasting 'election' data in a SMB domain you can just about forget
about the 50% too. As you know, with default settings, a TCP/IP stack that sees
50% collisions will slow down considerably, thus any imbalance in the
distribution will be immediately amplified.

Everyone knows download cases where one has maybe 3 parallel connections to the
same server, and they almost invariably become unbalanced, with one using up
the bulk of the bandwidth, and the others lagging behind more or less slowly.

Peter

2005\12\18@185021 by Gerhard Fiedler

picon face
Peter wrote:

>> faster than 512kbyte/sec DSL speed.
>
> That was 512 bbit/sec of course.

Or maybe 512 kbit/s ? (Not that nobody would've know this :)

Gerhard

2005\12\18@190207 by Gerhard Fiedler

picon face
Howard Winter wrote:

> If they get real experts, rather than equipment salesmen or bluffers,
> then they will tell them what you are trying to say, because that is the
> real situation.

If they get somebody, they should get somebody who guarantees a certain
well-defined performance level. No performance, no money... :)

Gerhard

2005\12\19@032903 by Nate Duehr

face
flavicon
face
On Sunday 18 December 2005 08:28 am, Howard Winter wrote:

> Yes, in the USA it's three, since that is the number of
> non-overlapping channels (In Europe & Japan it's more like 4 - I have
> no idea why the US chopped off the top few channels).  So the best
> case you could hope for is 3 x 40 users, so in an ideal situation
> each user would get 1/40th of 54Mb/s, which a rough calculation shows
> would mean a bit over a minute and a half to get 6MB to everyone.
> But of course the collisions problem makes this far from ideal, and I
> think you could reckon on at least 5 minutes in a real situation -
> maybe much longer.

I had to go dig some more and realized that the standard "Channels 1,6
and 11 don't overlap" isn't really truly the "whole picture".

Surprisingly, the best source of a good human-readable (not slogging
through the standards) discussion of this, I found at Wikipedia.  If
you trust the wiki after all the latest complaints of inaccurate
information in it... that is...

http://en.wikipedia.org/wiki/IEEE_802.11

1,6 and 11... DO overlap.  However, I can say that in practice, a friend
designed a number of 802.11b networks in warehouses for hand-held
bar-code scanners, and they worked, using the generalized assumption
that 1, 6 and 11 were the "non-interfering" channels to design the
network around.

The first deployment had channels scattered all over the place and
poor-placement of the access points.  His redesign move the AP's
appropriate distances away from one-another, and as much masking by the
steel pillar structures in the building of same-channel access points
as possible by "hiding" them from one-another, as much as possible.  It
also included adding directional antennas to get the signals to go
straight down toward the warehouse floor from the AP's mounted high
overhead, as possible.  

The overall re-design and new antennas, good quality co-ax to run from
the AP's the very short distance to the new panel antennas, and the
costs of flying him all over the place to clean them up and sniff for
problems, cost the company a small fortune.  

Their original vendor of the hand-held scanner equipment assured the
higher-ups at the company that the system was literally "Plug and
Play"... plop a bunch of AP's on the ceiling over every row and just
let the AP's have a free-for all, and the hand-held units would "just
work".  Ha.  They sold a lot of bar-code scanners that day.

> If they get real experts, rather than equipment salesmen or bluffers,
> then they will tell them what you are trying to say, because that is
> the real situation.

Show these "experts" the wikipedia link above, and see if they agree
with the throughput estimates and the notes about how all of these
network types fall back to lower rates as interference increases.   Ask
them how they're going to guarantee that the interference levels will
remain low enough for full data-rate use of the network in each covered
area.  Ask them if they've done an RF site-survey of your buildings.  
Get them on the defensive in front of the management, using solid
engineering questions and practices.

There's another aspect to this.  If you can distance yourself completely
from this project, do.  Real wireless networking experts wouldn't
suffer the fools who want to do this application wirelessly for very
long.  They chose the "yes men", and the people that are listening to
them truly deserve the pandemonium they've requested, if you've
adequately warned them and shown them engineering documentation to back
it up.

You may also think about keeping 802.11a in your hip pocket as an almost
usable solution to your problem... it's still not quite "right" but
AP's with 802.11a/b/g support and cards that do all three just MIGHT
get you enough bandwidth to make the application "liveable" if you can
convince the higher-ups that the higher costs (typically triple or
more) for tri-mode AP's is worth it.  

Also remember that the unlicensed segment of the 2.4 GHz ISM band is  
LITTERED with other devices... a student with a Bluetooth headset in
one of your classrooms becomes a HUGE network problem for your system
if it's 802.11b/g based.  Hell, a badly leaking microwave oven in the
teacher's lounge would too.

Good luck with it... sounds like it's going to get a lot more expensive
before it really works.

--
Nate Duehr, @spam@nateKILLspamspamnatetech.com

2005\12\19@033206 by Nate Duehr

face
flavicon
face
On Sunday 18 December 2005 01:13 pm, Peter wrote:
> samba/http/ftp servers each, then it would all boil down to using one
> commercial low cost 16-way router (or USB wireless adapter) per
> server to serve 120 people. The local network (server to router to

Danger: USB-to-Ethernet dongles typically won't run 10Mb/s reliably.

--
Nate Duehr, KILLspamnateKILLspamspamnatetech.com

2005\12\19@093728 by Mike Hord

picon face
> I had to go dig some more and realized that the standard "Channels 1,6
> and 11 don't overlap" isn't really truly the "whole picture".

> 1,6 and 11... DO overlap.  However, I can say that in practice...

I recall seeing a statement somewhere that, in a "lab setting", a
channel 1 file transfer slows down when a channel 11 transfer
begins.

Of course they "overlap".  Most analog signals do, to some degree.
That nth order Bessel filter you sweated over for two days to get
a really good 1 kHz cutoff still lets some 1.1 kHz signal through.
Attenuation is the key.  The 802.11 spec provides minimum
attenuations for signals at a particular distance from the center
frequency- something like -30dB for 2 channels away and -80dB
for 5 channels away.

> The first deployment had channels scattered all over the place and
> poor-placement of the access points.  His redesign move the AP's
> appropriate distances away from one-another, and as much masking by the
> steel pillar structures in the building of same-channel access points
> as possible by "hiding" them from one-another, as much as possible.  It
> also included adding directional antennas to get the signals to go
> straight down toward the warehouse floor from the AP's mounted high
> overhead, as possible.

This is what I want to do.  The classroom in question is perhaps 30x50
feet.  I'd like to put one WAP on each side of the room, with cantenna
type highly directional antennae, and turn their power WAY down.  That
SHOULD give enough attenuation that a user in the middle of the room
wouldn't see either side's WAP.  Put those on channel 6, then put a
channel one in the middle front and an channel 11 in the middle rear.

The bottom line is, I need to experiment some, and play around with it
and generally see what I can see.  The problem is, no one with the K
to back up my developing a good solution is willing to do so- they
just want to call in the "experts".

> Show these "experts" the wikipedia link above, and see if they agree
> with the throughput estimates and the notes about how all of these
> network types fall back to lower rates as interference increases.   Ask
> them how they're going to guarantee that the interference levels will
> remain low enough for full data-rate use of the network in each covered
> area.  Ask them if they've done an RF site-survey of your buildings.
> Get them on the defensive in front of the management, using solid
> engineering questions and practices.

The real, huge, almost insurmountable problem with this is University
bureaucracy.  If you've never worked from within a University, it can be
very frustrating at times.  At some point in the recent past, someone
heard the figure "no more than 30 users per WAP", and in their
non-technical minds, there is now an invisible wall at 30 users below
which things work swimmingly and above which they just flat fail.  Sigh.

> There's another aspect to this.  If you can distance yourself completely
> from this project, do.  Real wireless networking experts wouldn't
> suffer the fools who want to do this application wirelessly for very
> long.
> Good luck with it... sounds like it's going to get a lot more expensive
> before it really works.

Ayup.  But it ain't my money, and it likely won't be my problem much
longer, since I'm already shopping for a new job.  Actually, I guess it
IS my money, since my wife has already spent over $100k in tuition
at this institution...

Mike H.

More... (looser matching)
- Last day of these posts
- In 2005 , 2006 only
- Today
- New search...