Hi,
On Fri, Jun 08, 2001 at 02:35:45PM -0700, Bill Woodcock wrote:
> > But this is not a good long-term strategy. Annoy technicians with such
> > things (how much is there for Cisco to gain if I have to purchase a 1603R
> > because my trusty 2503 isn't supported anymore?)
>
> Ahem. I think the _correct_ answer is to buy a 1751, which comes with
> 128mb of RAM, 32mb of flash, and a real CPU. Why do people insist on
> buying dinosaurs?
Because they *do their job*.
It's not rocket science to pump IP packets through an ISDN dialup or
leased line, or through an T1/E1 link. You don't need a fancy RISC cpu to
do that.
The 1600 series is a nice design. They do not need so much space as the
1720/1750, come with the common interfaces on-board (so no need to pay
extra for "have a slot and get the card for it" if all you is Ethernet &
ISDN, or Ethernet & Serial), and are fast enough for what they are
supposed to do - connect small companies to the Internet.
I could turn around the question: why is Cisco still happily selling 1600s
(and 2500s)?
> I know the PLM for that space, and believe me, Cisco
> can't understand why people insist on continuing to order routers that are
> nearly a _decade old_.
Well - we don't purchase new 2500s. But we have a few of them, and they
just work, so why throw them away? (We *did* throw out the 7000s)
But we still order 1603s. Because they fulfill the needs of our
small customers. If the customer needs VPN performance, we'll sell him a
1720/1750, plus a crypto module, if needed. But why pay the surplus if he
doesn't need it?
One benefit of a 2500 over a 1720 is that the 2500 is 19" rack mountable,
which *is* a benefit for commercial customer. The 1720/50 is just ugly.
(Yes, the logical solution is to get a 2600 then. What we do, if we
want to purchase a 19" low-end router.)
> I mean, come on, folks, these are _68030s_. If
> you had a monitor and a keyboard on something that old, I don't think
> you'd be calling it "trusty", I think you'd have thrown it out the window
> in frustration, oh, say, six or seven years ago.
So? My home news server is a Sparc IPC. It doesn't have a monitor, it
doesn't need one - it does news serving, and DHCP, and does this well,
since a few years now. Why should I upgrade something that isn't broken
(and yes, I can put NetBSD on it and it will happily do IPv6)?
> Stop beating a dead horse and let Cisco move on. How can people
> complain that they're not able to beat Juniper on modern stuff, and then
> simultaneously whine that they're not adequately supporting legacy crap
> from the _previous century_?
Supporting "legacy crap" isn't very hard if the architecture is basically
the same throughout most of the older and newer boxes anyway ("shared memory
architecture", to quote from the IOS architecture book I'm just reading).
But this is just not being objective. The hardware design for a "beat
Juniper" box has to be very very different from the hardware design for a
"be cheap enough so that a SoHo customer can afford it" box. And the
1600/2500/2600s fulfill the lower end needs perfectly.
The basic problem here (IOS images getting bigger and bigger and needing
even more memory and flash) won't be solved by throwing away all 68k-based
routers, though. The only thing that will help here *are* custom linkable
images. Which isn't so "new" or "rocket science" either - Unix kernels
do it since at least a decade, and with good success.
gert
--
USENET is *not* the non-clickable part of WWW!
//www.muc.de/~gert/
Gert Doering - Munich, Germany gert@greenie.muc.de
fax: +49-89-35655025 gert.doering@physik.tu-muenchen.de
This archive was generated by hypermail 2b29 : Sun Aug 04 2002 - 04:12:40 EDT