From: Larson, Chris (CLarson@usaid.gov)
Date: Wed Feb 19 2003 - 17:53:32 GMT-3
Why even have a primary and secondary connections? Why not have equal cost
paths to each hub, get a couple large SANS, store all the data on the SAN
and replicate the SANS between hubs. Now your sites are load balanced and
fully redundant and load balanced.
I am only half kidding as I did not hear all those things in your post. I
just wanted to throw them in. The only thing with this is the pipe needed
for large amounts of data replication can be prohibitively expensive.
Although if you don't have too much data, maybe not.
> -----Original Message-----
> From: Chuck Church [SMTP:ccie8776@rochester.rr.com]
> Sent: Wednesday, February 19, 2003 1:48 PM
> To: Michael Snyder; ccielab@groupstudy.com
> Subject: Re: OSPF for 400+ Locations
>
> That's exactly what I had in mind. Never heard it called the 'hammock
> topology' though. You should copyright that one. I like it!
>
> Chuck Church
> CCIE #8776, MCNE, MCSE
>
>
> ----- Original Message -----
> From: "Michael Snyder" <msnyder@revolutioncomputer.com>
> To: <ccielab@groupstudy.com>
> Sent: Wednesday, February 19, 2003 11:45 AM
> Subject: FW: OSPF for 400+ Locations
>
>
> > I like hammock design for these things.
> >
> >
> > --(site 1)--
> > / \
> > ---(site 2)---
> > / \
> > (East Coast NOC)----(Site 3)-----(West Coast NOC)
> > \ /
> > ---(Site 4)--/
> > \ /
> > -(Site 5)-
> >
> > Each site has two frame-relay pvc's, one going to each NOC.
> >
> > Not necessary to have high cir's on both site pvcs thought. Use ospf
> > delay to select the primary pvc for each office. The backup pvc can
> > have zero cir.
> >
> > The NOC's have a fat pipe between them.
> >
> > It would be hard to knock down such a redundant design.
> >
> > BTW, the fat pipe between nocs and the main router loopbacks are
> > assigned to area 0 if you are using ospf.
> >
> > -----Original Message-----
> > From: nobody@groupstudy.com [mailto:nobody@groupstudy.com] On Behalf Of
> > Chuck Church
> > Sent: Tuesday, February 18, 2003 7:37 PM
> > To: Aaron Woody; mohammed@sulafsolutions.com; ccielab@groupstudy.com
> > Subject: Re: OSPF for 400+ Locations
> >
> > Aaron,
> >
> > With as large as this is getting, you really want to do this right
> > from
> > the start. Has the company given any thought to disaster recovery?
> > With a
> > thousand remote sites, this can't be a small company, so a circuit loss
> > at
> > the main site could be horrific. Backup PVCs to another location are
> > the
> > norm for a network of this size. But it's nothing that OSPF can't
> > handle.
> > Just design it right, and maybe you'll get a big bonus for a good job!
> >
> > Chuck Church
> > CCIE #8776, MCNE, MCSE
> >
> >
> > ----- Original Message -----
> > From: "Aaron Woody" <awoody@columbus.rr.com>
> > To: <mohammed@sulafsolutions.com>; <ccielab@groupstudy.com>
> > Sent: Tuesday, February 18, 2003 5:50 PM
> > Subject: RE: OSPF for 400+ Locations
> >
> >
> > > I just found out today they will be growing to 1000 sites and I will
> > still
> > > have to design dial backup solution.
> > >
> > > Aaron
> > >
> > > -----Original Message-----
> > > From: Mohammed Al-zubi [mailto:mohammed@sulafsolutions.com]
> > > Sent: Tuesday, February 18, 2003 3:40 AM
> > > To: 'Aaron Woody'; ccielab@groupstudy.com
> > > Subject: RE: OSPF for 400+ Locations
> > >
> > >
> > > Aaron,
> > > I've seen similar configurations, specially at banks (ATMs running
> > FR/OSPF).
> > > to be honest, if there are no networks behind the spoke routers (I
> > think
> > > there are none because you were going to configure them as Totally
> > stubby)
> > I
> > > would just implement static routes, you would just have to put a
> > default
> > at
> > > the remote sites, and 400 statics at the hub, its MUCH less work and
> > > headache, and this way when you roll it out you have less
> > configuration to
> > > deal with. If you still need these routes to appear in the OSPF
> > domain
> > > beyond the hub, then redistribute it back to OSPF at the hub with a
> > summary
> > > address, so 400 networks would show up in your site routing table as
> > one
> > > route, and you just simplified the configuration of 400 sites.
> > >
> > >
> > > Thanks,
> > > Mohammed
> > >
> > > <><><><><><><><><><><><><><><><><><><><>
> > > Mohammed Al-Zubi
> > > VP Professional Services
> > > 24 Werner Ave. #21
> > > Daly City, CA 94014
> > > Tel: (650) 438-6384
> > > Fax: (720) 293-4897
> > > Email: mohammed@sulafsolutions.com
> > > Web: www.sulafsolutions.com
> > >
> > >
> > > -----Original Message-----
> > > From: nobody@groupstudy.com [mailto:nobody@groupstudy.com]On Behalf Of
> > > Aaron Woody
> > > Sent: Tuesday, February 11, 2003 3:21 PM
> > > To: ccielab@groupstudy.com
> > > Subject: OSPF for 400+ Locations
> > >
> > >
> > > I have experience with OSPF but I am looking for suggestions on how to
> > > implement OSPF in a Frame-Relay Hub/Spoke topology for 400+ locations.
> > Each
> > > location only needs to know about the host through a default. My first
> > idea
> > > is to have a separate area for each location and make it a totally
> > stubby
> > > area. Is there a better way. My concern is that there will be 400+
> > areas
> > in
> > > the OSPF Database at the host. The host will be a Cisco 3745. The
> > remotes
> > > will all be Cisco 1751.
> > >
> > > Thanks!
> > >
> > > Aaron
> > >
> > > [GroupStudy.com removed an attachment of type application/ms-tnef
> > which
> > had
> > > a name of winmail.dat]
> > > .
This archive was generated by hypermail 2.1.4 : Sat Mar 01 2003 - 11:06:32 GMT-3