From: Simcha Blatter (SBlatter@xxxxxxxxx)
Date: Fri Aug 11 2000 - 12:00:08 GMT-3
Based on you original Diagram, Cisco currently recommends setting each of
the links between access and distribution as a standard VLAN connection. NO
Trunking. Each of the access layer switches will be configured in VTP
Transparent mode. This means, as in you diagram, that every access layer
switch in the wiring closets is its own subnet.
With this configuration in place the complexities and slow convergence of
spanning tree are eliminated.
HSRP with interface tracking is activated on the distribution switches so
that HSRP failover handles any failures. Routing Protocols between
Distribution and core converge quickly in the event of any failures between
distribution and core.
Load Balancing at the access layer is achieved by using MHSRP and balancing
the default gateways of network nodes between the 2 uplink vlan routed
connections.
This is the recommended method today by Cisco.
Please read the following Cisco White Papers for more info:
Gigabit Campus Network Design -- Principles and Architecture
http://www.cisco.com/cpropart/sync-src/ccstcp/cc/sol/mkt/ent/cmps/gcnd_wp.ht
m
Gigabit Campus Design -- Configuration and Recovery Analysis
http://www.cisco.com/cpropart/sync-src/ccstcp/cc/sol/mkt/ent/cmps/camp_wp.ht
m
With this design you do lose the flexibility of having network nodes be on
the same vlan across different wiring closets that have their own vlan
subnet.
Thanks,
Simcha
*****************************************************************
Simcha Blatter, Systems Architect
Dimension Data ISG
CCIE in progress, CCDP, CCNP, EMCess, MCSE, MCNE
*****************************************************************
-----Original Message-----
From: Richard Mott [mailto:richpmott@hotmail.com]
Sent: Friday, August 11, 2000 10:30 AM
To: philip.lai@ncr.com; ccielab@groupstudy.com
Subject: Re: CCIE Design Candidate , Pls read.
This looks like alot of equipment for what seems to be a network of about
100 users. (Someone is making a large commission here)
Gotchas:
At this time the 4000 series only suports 802.1q trunking.
(possible issues with 2924M which I believe are ISL only)
If you lose a distribution switch you isolate users and servers connected to
that switch.
If you lose a core switch you still isolate users and servers connected to
that switch.
Network design is like to tuning sound equipment: It's subjective.
Each person has there own design philosophies. (Which I think is going to
make the CCIE Design tough to pass)
Good luck
Rich Mott
CCIE #5234 (R&S)(ISP/Dial)
>From: Philip Lai <philip.lai@ncr.com>
>Reply-To: Philip Lai <philip.lai@ncr.com>
>To: ccielab@groupstudy.com
>CC: Philip Lai <philip.lai@ncr.com>
>Subject: CCIE Design Candidate , Pls read.
>Date: Thu, 10 Aug 2000 21:22:43 +0800
>
>Network Design issue.
>
> 6500(L3)- 6500 (L3) Core
> / \
> 4908G(L3) 4908G(L3) Distribution
> / \ / \
>2924M 2924M 2924M 2924M Access
> A B C D
>Vlan1 Vlan2 Vlan3 Vlan4
>
>Server Farm will be directly attached to the Core. There will be some
>departmental servers in zone A for both zone A and B users.
>
>My friend has designed a switching network like this and claim that it will
>have better performance than using 3508G (L2) in the distribution layer.
>Since the 4908G has the L3 routing feature, when zone A transfer packet to
>zone B, that traffic will pass only the 4908G but not the 6500 (Core).
>Thus,
>the bandwidth from the 6500 to 4908G can be saved.
>
>My questions:
>1. If both the Core and Distribution switch are routing enabled, that
>switches has to be the VTP server. ie: 4 indentical Vlan database are keep
>by each switches (Core & Dist.)
>2. Does it meaningful to configure trunk mode between all switches in order
>to pass the Vlan information ? If all 2924M are in VTP clients mode, which
>switches should be the VTP server ? and why?
>3. A crazy idea is to think about the L3 switches as a individual router.
>Then apply OSPF between to each one and enable routing. Will the network be
>quite slow for this design, compared with the standard design.-- Core
>(Routing)--Dis(Switching)---Access(Switching) ?
>
>4. In fact, I recommend the stand design using 3508G for dist. But could
>you
>tell me the adv and Disadv for using this and the design above ?
>
>Thank you very much for every reply
>
>Philip Lai
>
>
This archive was generated by hypermail 2.1.4 : Thu Jun 13 2002 - 08:24:24 GMT-3