Do you have:

A:  One office with a DSL router and cable router?

B:  Two offices, one with a DSL router and the other with a cable
router?

You might wonder why anybody would bother wiht a T-1 as it doesn't
seem any faster, and it is so much more expensive.  The reason is that
you can get service level agreements for T-1 circuits, and the telcos
are willing to do it becuase they've had the command, control and
monitoring infrastructure in place for decades.  Cable and DSL are
essentially best effort media.  If you are lucky, your ISP may do SNMP
monitoring of your cable or DSL box, and might call you when it drops
off their monitoring.  In the case of T-1, the infrastructure itself
is instrumented all along the span, so problems may be detected almost
immediately and [hiopefully!] dealt with quickly.

If you have situation A, you should talk to your service provider to
find out what sort of equipment is required.  T1 circuits come in a
variety of flavors, so you need to provision both the CSU/DSU and the
router correctly.  The CSU/DSU will accomodate parameters of the local
loop such as build-out capacitance compensation, etc.  Luckily, much
of the CSU/DSU's work is done automagically in modern devices.  The
T-1 circuit may be provisionied as B8ZS or it could be AMI.  It
depends on how the T-1 span is delivered to you, what you get.  Try to
look at the installation work order, and if you see anything about the
T-1 being provisioned as AMI, insist on having it upgraded to a B8ZS
span.  The router's serial port driver has to know whether it is
trying to sync to AMI or B8ZS - some routers can auto detect, but you
should insist on knwoing.  Beyond the low level, a T-1 may be
provisioned as ESF.  Again, this is a provisioning option on the
router.  The other thing you need to know is if you are buying a DS-1
T1 that is giving you a 1.536 megabit pipe to your ISP and TCP/IP
frames come out, or are you getting Frame-Relay service.  If you have
Frame-Relay, this is a higher order link control protocol - sort of
like X.25 Lite.  Frame relay allows the telco to virutalize your
circuit and multiplex your traffic on higher order circuits inside
thier network.  Frame-Relay isn't terribly attractive to you as a
customer unless you get a discount for agreeing to take your data that
way.  Where Frame-Relay is handy is if you have a hub and spoke type
office network.  You can use 64K Frame circuits to your satellite
offices, then take all the virtual circuits to your satellites over a
single T-1 Frame link to your hub.

I'd go with equipment that is likely to be known by a lot of network
professionals.  You can get corporate grade routers that have
integrated CSU/DSU cards from several manufacturers.  If you know what
you are doing, you can get a Cisco 2600 series router with a T-1 WIC
CSU/DSU card off eBay for a couple hundred dollars.  You want to be
sure to get both the router and the WIC together, if possible.  I'd
recommend  a minumum 8 meg flash and 32 meg RAM configuration on a
2600 - and if you think you'll add WICs, then you should try to get
the seller to send it to you with at least IOS 12.3(16a) installed.
The older IOSes don't support the WIC expansion cage for the left
slot.

If your service provider is willing to sell you a router, that is
probably the best way to go, especially if you can get them to agree
to configure the router for you as part of the deal.  If you don't
know your telecom and networking technologies pretty well, setting up
a T-1 can be a pretty daunting task.

Situation B is similar.  The only difference is that you are trying to
connect two parts of your private netowrk that happen to be separated
by a chunk of distance.  The important thing here is to make sure the
T-1 is configured the same on both ends.  You can run IP un-numberd on
the serial link and set up bridging on the routers.  No IP device need
be aware that some traffic is being bridged.  I belive in the case of
Cisco routers, if you enable bridging, you get the benefits of
broadcast isolation, and the routers will automatically proxy-arp fMAC
addresses of devices on the distant side of the network for TCP/IP
address resolution.  If you need to bridge Appletalk or Novell IPX/
SPX, you'll need  some additional configuration statements to proxy
traffic correctly.  For instance, you can enable your router to answer
Netware SAP rquests on behalf of servers on the distant side of the
network.

Situation B does not reuquire an ISP to get involved.  You lease the
T-1 from end-to-end.  The important thing is to make sure both ends
are provisioned the same.  If Frame-Relay is involved, it may be
possible to have it autoconfigure.  Again, you can fish for routers on
eBay if you know what you're doing, and are familiar with the
equipment.  Cisco 2600s with the WIC installed and at IOS 12.3(16a) or
better would be a good choice.  If you are in Stiuation B, the telco
might be willing (probably for $$$) to set up and configure the
routers.

Of course, there are lots of hardware choices.  For instance you can
probably make a router form a Soekris box and I belive have a CSU/DSU
daughter card right in the box.  The only reason I brought up Cisco is
that is is relatively easy to find somebody that's seen a Cisco router
and knows how to configure it.  If you roll your own, you'll  probably
be THE support person for it forever (nights and week-ends included!).

Bill

On Jun 26, 4:14 pm, Richard Clemens <rlclem...@clevelandtrack.com>
wrote: