Re: [j-nsp] OLIVE install

From: Jeff Bartig (jeffb@doit.wisc.edu)
Date: Mon Feb 19 2001 - 13:54:50 EST


On Mon, Feb 19, 2001 at 10:36:11AM -0500, Daniel L. Golding wrote:
> You could buy some M5's with Fast Ethernet PICs.
>
> Daniel Golding NetRail,Inc.

Hi Dan,

If we decide to start buying Junipers, we will probably end up doing
something like that, although it probably won't be "some". More likely
it would be one initially.

Historically, we have been purchasing Cisco routers for our network.
We currently have two GSR's purchased in 1998, before the Juniper
was really an option. Over the next year, we need to buy three
more high-end routers (GSR or Juniper). Maybe a total of five if
a special funding request we made gets passed by the State of
Wisconsin Legislature. Our budget is quite tight and it is unlikely
I would get a request for several lab routers approved.

With the need to purchase additional high-end routers, I felt it
was wise to evaluate our options. One small piece of my evaluation
is the cost of lab equipment to simulate parts of our network and
for staff training. For this particular part of the evaluation,
Cisco is definitely going to come out ahead. First, they have the
unfair advantage that we have been using their products for years,
so our lab is already well stocked with routers that have been
decommissioned, spare routers that would have to leave in case of
failure, and routers on loan or donated by vendors. Cisco also
has an advantage that they have 12.0S code available for most of
these routers in my lab. Note that I currently have no GSR's in
my lab, but I am able to get by with what I have for most lab work.
If I do need a GSR for a specific test or evaluation, I can always
ask Cisco to loan me one.

Since Juniper would be a new addition to our network, we would be
starting with nothing in the lab. I had been looking at the Olive
as a possible way to jump start my lab environment with an inexpensive
investment in a couple of PC's. This will not be possible. Instead,
I will probably be looking at an M5 or more likely an M10 (same as
we would be deploying in production, so I could also use sparing
as a cost justification). Looking at the pricing for an M5, I
come up with:

        M5BASE-AC $10,000
        RE-333-256 20,000
        JUNOS 10,000
        PE-TUNNEL 10,000
        PE-1GE-SX 20,000 (I don't have a price for the FastEther)
                        --------
        Total $70,000 list (+$10,000 list for M10 rather than M5)

Even discounted, this is pretty expensive. If we go with the
Juniper, it will be something we need to do. When we started buying
Cisco routers, we didn't have much of a lab either.

Like I said earlier, though, this is only one small aspect of the
overall evaluation. They didn't do well on this part, but things
look very good for them on other aspects of the evaluation. All
of these things need to be taken into account before we make a
decision. It may be worth overlooking lab issues in the short-term
in order to end up with a better network in the long-term.

Maybe we can find some time to chat today or tomorrow about your
thoughts of the Juniper. It would provide good input for my
evaluation.

Jeff

-- 
Jeff Bartig                  |  University of Wisconsin - Madison
1210 W Dayton, Rm B263       |  Division of Information Technology
Work Phone: (608) 262-8336   |  Network Engineering Technology
E-Mail: jeffb@doit.wisc.edu  | 



This archive was generated by hypermail 2b29 : Mon Aug 05 2002 - 10:42:40 EDT