Design Queries

ling20125 · ‎05-31-2010

Hi everyone,

I'm Angela, and this is my first post here in CSC, got some design implementations problems. I am reading Cisco LAN switching by Clark and Hamilton, published in 1999. Not the greatest book for latest technology, but I do find a lot of design tips to be very helpful. Here we go:

1- What is your ideal OSPF area versus IP addressing plan?

2- The book I mentioned above suggested using L3 switches as core switches, I've also seen some L2 cores. Please comment on the pros and cons of each

Got some technical questions as well:

1- Does software-based broadcast suppression is still around? If so, please tell me how to configure it

2- What really triggers a switch port to enter listening state (from blocking)?

3- Here is a point from the book (not exact wording): "A TCN BPDU is generated when a port moves to forwarding state and the switch has a Designated port", why does a Designated port has to be around for a TCN BPDU to be generated because a port moves from blocking to forwarding?

Thanks for all comments,

A

vmiller · ‎05-31-2010

regarding question 1. Don't tie the two concepts together too closely.

OSPF area design should be done with the idea of keeping a magageable and robust area0, with the attached areas summarizing correctly.

I'd just insure that I had enough address space for my areas to keep summarization tidy, the more contiguous, the better,

Having done more than one campus design, I can testify that the redundant layer 3 design is in my opinion the only way to go. THis also speaks to

question one, in that correct summarization combined with well designed Layer 3 redundancy at the core equals optimal reconvergence.

Leo Laohoo · ‎05-31-2010

The book I mentioned above suggested using L3 switches as core switches, I've also seen some L2 cores. Please comment on the pros and cons of each

Times has changed since 1999 ... alot! One of the benefits of L3 switches as core is that it can do routing. It's beneficial if you are separating your networks per subnet. However, this is not a mandatory. It depends entirely on the size of your network. If you have a network of say, a few people (20-50) and it's not really complicated (a few simple servers, PCs, printers) then getting a L2 for core will do well. I mean L2 switches are more affordable compared to L3 obviously but you don't use this as a reason at all. I've worked in an organization where we use L2 switches for just about everything and we had only one VLAN. I'll tell you this, one broadcast storm can bring your network to it's knees very quickly.

lamav · ‎05-31-2010

Leo makes a very valid point, Angela.

Reading that book is going to give you a lot of information that has been deprecated as a result of technological advancements.

For example, running L3 in the core is may favorite approach in an enterprise environment, mainly because I dont want layer 2 loop issues in my core. That having been said, with tthe advent of VSS and multi chassis etherchannel, STP is no longer needed as a loop removal mechanism because there are no loops.

You should read Cisco Network Design Guides instead.

http://www.cisco.com/en/US/netsol/ns742/networking_solutions_program_category_home.html

Victor

ling20125 · ‎06-01-2010

Thanks for the tip, can someone help with some of my technical questions?

Reg