WAN Link

I am horribly new to the whole wireless thing (PACKETS ARE SUPPOSED
TO TRAVEL OVER COPPER!!!), and am a bit overwhelmed with the available
information. I need to set up a point to point WAN link, and would
appreciate pointers in the right direction - I don't expect anyone
to "do my homework" for me. The requirements for the link are
pretty simple.

It has to be reasonably priced, as it'll only be operating for
approximately a month during a facility move, so there's *no*
*way* I'm going to get a real budget. A simple bridging setup is
desired - routing would complicate network migration more than
necessary. The two buildings are approximately 1/4 mile apart,
I have rooftop access to both, and there's lovely unobstructed
line of sight between them.

Any tips and/or usable resources would be greatly appreciated!

Thanks,
JS

On Tue, 15 Nov 2005 12:35:00 -0500, John Schmidt <(E-Mail Removed)> wrote:

>I am horribly new to the whole wireless thing (PACKETS ARE SUPPOSED
>TO TRAVEL OVER COPPER!!!),

Also over glass (fiber).

>and am a bit overwhelmed with the available
>information.

If it were easy, it would be no fun. Think of 802.11 style wireless
as nothing but bridging. Everything is done on layer 2. IP addresses
only appear to manage the various boxes.

>I need to set up a point to point WAN link, and would
>appreciate pointers in the right direction - I don't expect anyone
>to "do my homework" for me. The requirements for the link are
>pretty simple.

Assumption, the mother of all screwups.
What you're looking for is a "wireless bridge". More specifically, a
transparent bridge. This can be anything from a pair of cheapo
Linksys WAP54G or WRT54G (with alternative firmware) routers, to a
very expensive RedLine AN-50 wireless link that will do 48Mbits/sec.

http://www.redlinecommunications.com..._overview.html
There are also integrate radio/antenna solutions that might be worth
looking at:
http://www.winncom.com/products/category/WPP/list.html
However, I can't even begin to suggest hardware and sources until you
disclose some more numbers. See below.

>It has to be reasonably priced, as it'll only be operating for
>approximately a month during a facility move, so there's *no*
>*way* I'm going to get a real budget. A simple bridging setup is
>desired - routing would complicate network migration more than
>necessary. The two buildings are approximately 1/4 mile apart,
>I have rooftop access to both, and there's lovely unobstructed
>line of sight between them.
>
>Any tips and/or usable resources would be greatly appreciated!

I need a bit more info:
1. Are you SURE you have line of sight including the Fresnel zone?
2. How much thruput (mbits/sec) do you need to move?
3. Do you have power on the roof or will you need to run PoE (power
over ethernet)?
4. How many MAC addresses (total) need to be seen through the bridge.
5. Have you done a site survey to check if there will be any
interference? Are there any existing 2.4GHz systems nearby on either
rooftop or nearby buildings? (Use binoculars).
6. Approximate price limit for hardware. You're on your own for
installation, config and setup.

--
Jeff Liebermann (E-Mail Removed)
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558

John Schmidt wrote:

> I am horribly new to the whole wireless thing (PACKETS ARE SUPPOSED
> TO TRAVEL OVER COPPER!!!),

Really? I thought it was all magic :-)

> and am a bit overwhelmed with the available
> information. I need to set up a point to point WAN link, and would
> appreciate pointers in the right direction - I don't expect anyone
> to "do my homework" for me. The requirements for the link are
> pretty simple.
>
> It has to be reasonably priced, as it'll only be operating for
> approximately a month during a facility move, so there's *no*
> *way* I'm going to get a real budget. A simple bridging setup is
> desired - routing would complicate network migration more than
> necessary. The two buildings are approximately 1/4 mile apart,
> I have rooftop access to both, and there's lovely unobstructed
> line of sight between them.
>
Pretty well any wireless router that permits an external antenna to be
connected, and any wireless device that will act as a bridge (and take an
external antenna) to it will do the job.

I'd recommend something like my setup - two linksys WRT54Gs (< $70 ea), two
small FAB-corp dishes ($30-40 ea) with pigtails for the Linksys (< $30ea)
and sveasoft firmware ($20), except that you may have trouble finding
WRT54Gs that you can install the new firmware on (if the V4.x/V5.x WRT54s
can do bridging natively, they'll be fine). The whole thing should cost
under US$300 including your time.
--
derek

Jeff Liebermann wrote:
> On Tue, 15 Nov 2005 12:35:00 -0500, John Schmidt <(E-Mail Removed)> wrote:
>
>
>>I am horribly new to the whole wireless thing (PACKETS ARE SUPPOSED
>>TO TRAVEL OVER COPPER!!!),
>
>
> Also over glass (fiber).

Yeah, I forget about fiber. With the advent of GigE, I really don't
have a use for it anymore.

>
>>and am a bit overwhelmed with the available
>>information.
>
>
> If it were easy, it would be no fun. Think of 802.11 style wireless
> as nothing but bridging. Everything is done on layer 2. IP addresses
> only appear to manage the various boxes.

Which is precisely the scenario I'm looking for. Being forced into a
routed solution would mean a god-awful amount of extra work, given
my current network topology.

>>I need to set up a point to point WAN link, and would
>>appreciate pointers in the right direction - I don't expect anyone
>>to "do my homework" for me. The requirements for the link are
>>pretty simple.
>
>
> Assumption, the mother of all screwups.

Agreed - by "simple", though, I meant "not complex from a networking
standpoint".

> What you're looking for is a "wireless bridge". More specifically, a
> transparent bridge. This can be anything from a pair of cheapo
> Linksys WAP54G or WRT54G (with alternative firmware) routers, to a
> very expensive RedLine AN-50 wireless link that will do 48Mbits/sec.
>
> http://www.redlinecommunications.com..._overview.html
> There are also integrate radio/antenna solutions that might be worth
> looking at:
> http://www.winncom.com/products/category/WPP/list.html
> However, I can't even begin to suggest hardware and sources until you
> disclose some more numbers. See below.

Thanks for your patience. I'm so new at this I don't even know the
proper questions to ask.

<snip>

> I need a bit more info:
> 1. Are you SURE you have line of sight including the Fresnel zone?

Dead sure - that's the one thing I do know.

> 2. How much thruput (mbits/sec) do you need to move?

10 Mbs would make me ecstatic. I can get away with 2 or 3 Mbs, as long
as it's pretty much constantly available.

> 3. Do you have power on the roof or will you need to run PoE (power
> over ethernet)?

I have 110v AC readily available on both roofs.

> 4. How many MAC addresses (total) need to be seen through the bridge.

A maximum of 2 dozen - after migration is complete, there will be a lot
more, but I can tweak the schedule to minimize what will need to be
carried over the wireless link.

That question is a *great* example of how ignorant I am, BTW. I'm used
to dealing with powerful enough Cisco routers and switches that MAC
address count has simply never been an issue I've had to deal with.

> 5. Have you done a site survey to check if there will be any
> interference? Are there any existing 2.4GHz systems nearby on either
> rooftop or nearby buildings? (Use binoculars).

None of one roof, nor any in between. On the other roof, I'm not sure.
I know a bit about ham antennas, but I wouldn't know the difference
visually between a 2.4GHz and a microwave (or pretty much any other
type) of antenna.

I guess a chat with the owner of the new building is in order to find
out what the couple of antennas (antennae?) are.

> 6. Approximate price limit for hardware. You're on your own for
> installation, config and setup.

Price limit is tricky. The way my boss normally works for this kind
of stuff is I give her a close cost estimate for a given project, and
it's either accepted or rejected. I have to be a bit of a mind reader.

Again, I *really* appreciate the advice. If I knew just a bit more
about the subject, I'd know enough to be able to recognize a good
resource (web site, book) to answer my questions. But this is
absolutely virgin territory for me...

JS

On Tue, 15 Nov 2005 15:14:47 -0500, John Schmidt <(E-Mail Removed)> wrote:

>Agreed - by "simple", though, I meant "not complex from a networking
>standpoint".

Bridging is about as simple as you can get. As long as you aren't
trying to glue together two giant networks, bridging should work. The
bottom of the line wireless bridges will do 32 MAC addresses. The
better ones will do 256. The overpriced stuff will typically do 2048.

>10 Mbs would make me ecstatic. I can get away with 2 or 3 Mbs, as long
>as it's pretty much constantly available.

As a rule of thumb, you'll get slightly less than half your connection
speed. At 1500 ft (1/4 mile) with 802.11g, you'll get at least a
12Mbit/sec connection, which will yield about 6Mbit/sec thruput. It
will probably be higher but that really depends on the gain of the
antennas and the local RF noise pollution. I'll try to grind the
numbers to go faster later.

>I have 110v AC readily available on both roofs.

This is too easy. Can't you offer some kind of complication to add to
the challenge? Ideal installations like this just never seem to
happen to me. I NEVER seem to have AC power anywhere near where it's
needed. (Hint: Don't run sensitive electronics on the same power
that runs the elevator or rooftop HVAC motors. The power tends to be
really glitchy).

>> 4. How many MAC addresses (total) need to be seen through the bridge.

>A maximum of 2 dozen - after migration is complete, there will be a lot
>more, but I can tweak the schedule to minimize what will need to be
>carried over the wireless link.

24 MAC addresses is doable with literally anything you can buy.
However, note that this is the total number of MAC addresses for both
sides of the bridge.

>That question is a *great* example of how ignorant I am, BTW. I'm used
>to dealing with powerful enough Cisco routers and switches that MAC
>address count has simply never been an issue I've had to deal with.

I know the feeling. The first time I slammed into a 32 or 256 MAC
address limit, I was amazed that it could be so limited. However,
with bottom of the line cost cutting, even cheap RAM is considered
fair game for the bean counters. The bad news is that many of the
cheap bridges and routers do not recover gracefully from overflowing
tables. Some hang, some crash, some reboot, and some actually do the
right thing and impliment a least recently used algorithm and expire
the oldest MAC's.

>> 5. Have you done a site survey to check if there will be any
>> interference? Are there any existing 2.4GHz systems nearby on either
>> rooftop or nearby buildings? (Use binoculars).

>None of one roof, nor any in between. On the other roof, I'm not sure.

Also, check if there are any funny looking antennas on the rooftops
along the line of sight and beyond the target building. You may find
it expedient (or convenient) to mount the antenna in a window rather
than on the roof top. You'll pickup much less interference that way.
The idea is to see the other end of the link and not the entire city.
Unlike lower frequency land mobile installs, higher is not necessarily
better for point to point.

>I know a bit about ham antennas, but I wouldn't know the difference
>visually between a 2.4GHz and a microwave (or pretty much any other
>type) of antenna.

Just look for dish antennas. The 2.4Ghz variety are usually barbeque
grill types of dish antennas. The 5.7GHz variety are usually solid
dish antennas. See:
http://www.fab-corp.com
for typical antennas. The narrow 0.5ftx3ft fiberglass panel antennas
are usually for cellular or paging. There are some 2.4Ghz sector
antennas that look very much like cellular antannas but those are only
for WISP (Wireless ISP) service and are usually easy to discover with
a laptop access point scanner (Netstumbler for Windoze or Kismet for
Linux). If you're near a University, you may be in trouble as they
tend to deploy huge numbers of hot spots.

>I guess a chat with the owner of the new building is in order to find
>out what the couple of antennas (antennae?) are.

If you find the roof occupied with 2.4GHz hardware, all is not lost.
There are 3 non-overlapping channels (1, 6, and 11). Just pick the
one that's hopefully unoccupied. Also, you might wanna read:
| ftp://download.intel.com/business/bs...nt/hotspot.pdf
It's really made for hotspot deployment, but all the terms, technology
and buzzwords are relevent to anything that uses 802.11a/b/g. Well
worth a reading, methinks.

>> 6. Approximate price limit for hardware. You're on your own for
>> installation, config and setup.
>
>Price limit is tricky. The way my boss normally works for this kind
>of stuff is I give her a close cost estimate for a given project, and
>it's either accepted or rejected. I have to be a bit of a mind reader.

Well, I was hopeing for order of magnitude. Since this is allegedly
temporary, methinks maintaining resale value is also a consideration.
Therefore, I suggest an integrated unit with antenna, radio, and power
all in one package, with a minimum of configuration and installation
issues.

Tranzeo TR-6015 or TR-6019 point to point bridge:
| http://tranzeo.com/index.php?section..._section_id=44
for about $300/each end is methinks ideal, but possibly expensive.

If you want to do 5.6GHz instead, the TR-5a-21f will also work for a
bit more money:
| http://tranzeo.com/index.php?section..._section_id=47

At the low end of the scale, there's a pair of Linksys WAP54G bridge
radios for about $50/ea. Add two antennas of about 15dBi gain for
about $40/ea and two coax pigtails for about $15/ea. Total is
slightly over $200.

You can also use a pair of WRT54G radios, and install alternative
DD-WRT firmware to give it bridging capeabilities. However, this
solution requires considerable tinkering and will involve considerably
more effort. I think the WRT54G give 256 MAC addresses, while the
WAP54G only gives 32 MAC addresses.

You have one of the simplest installations possible. All you need to
do is configure the:
1. Channel numbers
2. SSID (system name)
3. MAC address of destination bridge for security.
4. Encryption level (WPA if possible) and pass phrase (20 chars
minimum).

That's it. There are some issues with proprietary enchancements
(Turbo, Super G, etc) but you probably won't run into those.

--
Jeff Liebermann (E-Mail Removed)
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558

Jeff Liebermann wrote:

< loads of great information snipped throughout >

> This is too easy. Can't you offer some kind of complication to add to
> the challenge?

The entire migration is going to be challenging enough. Try to
remain sane while co-ordinating moving the local loops for a
dozen or so data lines ranging in size from T1 to OC3.

I'm going to put one of my therapist's kids through college
by the time this is done...


> Ideal installations like this just never seem to
> happen to me. I NEVER seem to have AC power anywhere near where it's
> needed. (Hint: Don't run sensitive electronics on the same power
> that runs the elevator or rooftop HVAC motors. The power tends to be
> really glitchy).

Added bonus - it's filtered, clean, reliable power from UPSes.
I guess *something* had to go right.


> The bad news is that many of the
> cheap bridges and routers do not recover gracefully from overflowing
> tables. Some hang, some crash, some reboot, and some actually do the
> right thing and impliment a least recently used algorithm and expire
> the oldest MAC's.

The cynic in me thinks that behavior might be (at least a bit)
intentional. Vendors would rather sell high end gear than entry level.
That cynicism isn't solely directed at wifi vendors, but rather the
networking industry as a whole.

> Tranzeo TR-6015 or TR-6019 point to point bridge:
> | http://tranzeo.com/index.php?section..._section_id=44
> for about $300/each end is methinks ideal, but possibly expensive.

That looks like a *great* solution. Just a touch more than I wanted to
suggest spending, but the time savings from an all-in-one solution
should more than make up for it.

Thanks again, Jeff. You have been a _HUGE_ help.

JS

On Wed, 16 Nov 2005 10:00:48 -0500, John Schmidt <(E-Mail Removed)> wrote:

>The entire migration is going to be challenging enough. Try to
>remain sane while co-ordinating moving the local loops for a
>dozen or so data lines ranging in size from T1 to OC3.

Ouch. I've helped with minor moves and found that even if we get 99%
of the changes correct, the remaining oversights will burn most of the
next week. I learned that it's best to assume that literally
everything will need to be tweaked, tested, and documented. Trying to
get the telco monopoly to fix their provisioning on a weekend is
impossible.

What nearly killed me was physical access. We had done a successful
dry run on a 200 desk move. But on the day of the move, the high
command had decided that this would be a great time to take a vacation
so as not to get blamed if it failed. They took the keys and codes
with them and literally disappeared. I had to pick locks and do some
password bypass games on the network boxes. I got lucky and only set
off one alarm. Suggestion: Get keys and codes to the people that
need them BEFORE the move.

Also, T1 and even OC3 are considered commodity bandwidth by most telco
monopolies. Getting them to fix the inevitable errors in the
provisioning over a weekend is impossible. 3 days is about right. We
also covered ourselves on one move by anticipating that the T1 PRI,
the DID lines, and the rotarty would get mangled during the
transition. (Always happens). We had pre-arranged for a VoIP
provider to handle the company voice traffic. When the PRI didn't
work, I just cut over the phone system to the WAN connection. Internet
access was a bit constipated for a while, but it bought me the
necessary time to get the telco untangled.

>I'm going to put one of my therapist's kids through college
>by the time this is done...

Suggestion: Read some Dilbert cartoons and books before proceeding.
It will put you in the right frame of mind. Also, if you're in
charge, plan on practicing some sleep deprivation and practice living
off delivered junk food (pizza, donuts, etc).

>Added bonus - it's filtered, clean, reliable power from UPSes.
>I guess *something* had to go right.

On the roof? Surely these buildings must be IT heaven. I've had to
install ferroresonant transformers on rooftop power to keep the
building garbage out of the power. On one rooftop, I was blessed with
90VAC on one phase, and 160VAC on the other.

>The cynic in me thinks that behavior might be (at least a bit)
>intentional. Vendors would rather sell high end gear than entry level.
>That cynicism isn't solely directed at wifi vendors, but rather the
>networking industry as a whole.

I don't think so. The short product life of the bottom of line
wireless products do not lend themselves to long term quality and
testing. Many of the vendors involved literally do not have any high
end gear to promote.

>> Tranzeo TR-6015 or TR-6019 point to point bridge:
>> | http://tranzeo.com/index.php?section..._section_id=44
>> for about $300/each end is methinks ideal, but possibly expensive.
>
>That looks like a *great* solution. Just a touch more than I wanted to
>suggest spending, but the time savings from an all-in-one solution
>should more than make up for it.

Well, if the $600-$700 total price is too high, there are cheaper
versions with the same form factor.

YDI was purchased by Terrabeam which was recently bought by bankrupt
Proxim. The original product was called "Etherant" and is available
in versions I through III. There is also "Bridge in a box". They can
still be found at various wireless dealers and on eBay.
http://www.ydi.com/products/bridges/bridge-in-a-box.php
They are 802.11b only which will give you about 4Mbits/sec maximum
thruput. Make sure you get the wireless bridge and not the client or
access point version. Not great, but useable. I would spend the
money on 802.11g Tranzeo units as the cost savings for 802.11b only is
not very large.


--
Jeff Liebermann (E-Mail Removed)
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558