Karlnet
[Top] [All Lists]

[Karlnet] Super TurboCell Base Stations?

To: <karlnet@WISPNotes.com>
Subject: [Karlnet] Super TurboCell Base Stations?
From: "Brian H. Oak" <oakb@infosystems1.com>
Reply-to: karlnet@WISPNotes.com
Date: Sat, 14 Sep 2002 13:27:15 -0600
List-post: <mailto:karlnet@WISPNotes.com>
I am looking for some kind of super-duper TurboCell base station.  I
have tried products from several different manufacturers, but am still
at a loss.  Our current standard base station is the Agere COR-1100 or
Avaya COR-II, but I know I can do better when it comes to throughput.

I have used C-Spec RF-11+es in the past, and found them to be the
fastest thing available.  The C-Spec product has several
disadvantages, however:

}} Only 1 radio and 1 FE port, making base installs complex, ugly, and
expensive

}} Lack of support for newer KarlNet images

}} Inexpensive moving components (read: fan) can break down and ruin
base

}} Over-all reliability is very low, short MTBFs

}} Problem resolution takes a long time, sometimes forever (i.e. some
problems have never been solved, seem to have been forgotten by the
folks at C-Spec)

}} The cost of the RF-11+ is comparatively very high, although I would
be willing to pay that much in the absence of these other problems

Additionally, I have done extensive testing of the FE interfaces on
the RF-11+es: I found their 100Mbps Full-Duplex flow control to be
ineffective, forcing static configuration at 10Mbps Half-Duplex.  In a
way this makes sense, since the radio interface is limited to 11Mbps
nominal.  But the FE interface itself should handle flow/congestion
control so that very few frames are unanswered and dropped.  In all
fairness, this might be a problem with all FE interfaces on KarlNet
boxes, I haven't done such testing on the only other FE-available
boxes of which I know, the COR-1100s.

I have to wonder, though, if a FE interface with a chipset from one of
the premier NIC manufacturers (Intel, 3Com, etc.) wouldn't do a better
job -- and if more traffic I/O RAM should be used to allow more of a
traffic-shaping thing to happen in the store-and-forward switching
process.  That would, theoretically at least, make it so that no
traffic was lost, it would just be stored until it could be doled out
over the slower radio interface.  This is the kind of thing that
"real" routers do all the time, going from a 100Base-TX interface to a
1.544Mbps T1.

Anyway, I'm looking for a manufacturer who has built a box with the
high throughput I know can be achieved, without the major downsides
mentioned above.  Anyone know of one?

I'm thinking of a box, 19" rack-mountable (no more than 2RUs), running
one of the newer CPUs => 1GHz, lots of RAM, capable of supporting up
to four radio interfaces and multiple FE interfaces, all running the
KarlNet v4+ TurboCell software.  If I can't find someone making a box
like this, I might just start making them myself.  That is, of course,
assuming I can get someone at KarlNet to return my phone calls (no
success so far).  Is anybody else interested in such a product?

Thanks,

-Brian
____________________________________________________________
Brian H. Oak    CISSP?CCDP CCNP CCA NNCAS    Datawav-IS, LLC
Principal Engineer?        ??   109 N. Arthur Ave., Ste. 301
(208) 232-6860????????????????          Pocatello, ID? 83204
(208) 232-0185?fax????????????  ??  ??brian@infosystems1.com


<Prev in Thread] Current Thread [Next in Thread>