Cisco Support Community
Showing results for 
Search instead for 
Did you mean: 

Welcome to Cisco Support Community. We would love to have your feedback.

For an introduction to the new site, click here. If you'd prefer to explore, try our test area to get started. And see here for current known issues.

New Member

LAN Switching from access to core

I am evalulating a hospital LAN which has 3 access switches (144 ports) connected to each other with

fiber and then collapse on a single fiber run back to the core. I have calculated a 14 to 1 bandwith ratio.

14Gb access ports to 1 Gb fiber uplink port. What is acceptable? The users are traveling back to the core for all

applications and (there are anywhere from 3 to 5 wireless AP's connected to the access switches). There are 20 closets like the one described above which connect to an Extreme Alpine 3808.

Any help would be appreciated.



Cisco Employee

Re: LAN Switching from access to core

Hi Steve,

The answer, as with most design questions is "It depends." The area of focus should be the link with the smallest bandwidth. If you are not overrunning the most oversubscribed link during peak periods, then yes, it is acceptable. Of course, you'll want to factor any expected or planned growth into that.



*Please rate helpful posts.

Hall of Fame Super Silver

Re: LAN Switching from access to core

At first blush, it would appear that the core uplinks are way oversubscribed. However, before making any capital investment I'd recommend setting up some performance monitoring to see just how things look by gathering empirical data and then moving on to fact-based decision making. Invest $1000 or so in Solarwinds or other easy to use monitoring system (if you're handy with software and don't mind wokring under the hood, try one of the open source tools - mrtg or cacti). Once you have data, you can make better-informed decisions about architecture.

If end users have applications stored locally and data as well, things might be in not so bad shape. If they have big data sets and files on central servers, things are probably strained. If they really pull applications from servers (client software and all), then they are probably banging on your door. All of these scenarios might lead one to different architectural decisions (and attendant costs). That alpine swithc in the core is a bit long in the tooth and may be a bottleneck if you need to increase uplink quantities and throughput.

You say they have gigabit ports (10/100/1000?). How many of them are actually running at 1 Gbps? The APs are likely somewhat self-limiting due tho the shared bandwidth nature of the clients - unless you have high density client access and something like micro/pico cells.

Hope this helps. Please rate helpful posts.