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

Switch performance: bandwidth and fabric

Can someone explain me differences between these terms? Both are measured in Gbps.

  • Getting Started with LANs
New Member

Re: Switch performance: bandwidth and fabric


The amount of data (in bits) that can be transmitted per second over a particular network connection, or through a system.

Switch Fabric

The foundation of a router, interconnecting incoming data from ingress (input) ports to egress (output) ports. There are multiple switch fabric architectures with differing strengths in terms of bandwidth, throughput and delay.


New Member

Re: Switch performance: bandwidth and fabric

Can you explain this for Catalyst 3750E as example? In brochures I have seen:

- StackWise Plus for ease of use and resiliency with 64 Gbps of throughput;

- 128-Gbps switching fabric;

How many full-duplex 1Gb links (theoretically fully loaded in both directions) can be connected to a stack of several switches to achieve overburdening?

And how about 3750G, for witch I didn't find any throughput data, only 32Gb switch fabric.


Re: Switch performance: bandwidth and fabric

Not sure I understand your question. A Cisco 3750G switch with a 32Gb fabric can support line rate (bi-directional traffic) on all ports simultaneously. Obviously the theoretical maximum bandwidth through the switch is exactly 50% of the total of its switch ports because traffic is thru a switch, in and out. You can't send traffic in on more than 50% of the ports because the traffic wouldn't have anywhere to go.

All Cisco stackable switches made within the last 7 years or so have supported layer-2 line rate on all ports.

The better question to ask, however, is what is the layer-3 forwarding rate. In some cases, that number (measured in pps or packets per second) can be significantly less than the layer-2 theoretical forwarding rate, particularly with smaller (< 128 byte) packets.

From a slightly different angle, all these numbers change drastically when you 'stack' switches (3750) or cross-connect switches (2960/3560/etc) or use a chassis based solution (4500/6500, depending on supervisor and linecards). The uplinks to a core switch also can have a significant impact on performance.

The rule of thumb for good design is between 8 and 16 to 1 oversubscription at the edge (48 gig copper ports to a 4-port gig uplink bundle), no more than 4 to 1 oversubscription at the distribution layer, and absolutely *ZERO* oversubscription at the core and server farm layers.

I would always recommend you eliminate oversubscription at the distribution layer if at all possible in your design.

Keep in mind that statistically 300 users in an edge switch stack or chassis will not, with high probability, be transmitting or receiving data at line rate at the same time. In the distribution and particularly the core and server farm layers, this is not the same. There is a very high probability that hundreds or thousands of users could be requesting significant bandwidth from several servers simultaneously. This obviously is determined more by the type of industry and types of applications deployed, but is a good general rule.

New Member

Re: Switch performance: bandwidth and fabric

I am trying to calculate the “limits” for using the Catalyst 3750G stack as netwok core. This solution is designed by Cisco as best for a branch up to 200 users. For a LAN with 250-500 users Cisco recommend Catalyst 4500 as a core. I suppose this number is pure theoretic. I want to find out some practical guidance how to calculate realistic demands in a concrete situation.

For example I have a LAN with 250 with users and up to 8 servers, all of them with 1Gb Ethernet (this is not significant, it is easier to calculate). Let’s suppose there isn’t traffic between desktops. Some servers are more charged, some less.

As best practice, for 250 with 1:20 oversubscribing (recommended values are 1:15 to 1:20) I must have 12Gb downlink throughput. But there are only 8 servers and I suppose I don’t need 8G for them, all the more 12 Gb. So, how thick must be common downlink throughput for these 8 servers?

And let’s suppose I have more serves – 20:-). I suppose in this care I have to ensure this 12Gb throughput from edge switches to core. For most switches Cisco provide the switch fabric performance values. In this term how “big” must be the core to ensure these 12Gb downlinks:

- 12Gb switch fabric;

- 24Gb (because I have 1Gb full duplex);

- 48Gb (because core must process traffic from uplinks and “responding” one from servers)?

As far about my example with Catalyst 3750G. I meant a stack from one 3750G-12S, where all uplinks are connected and one 3750G-24T, where serves are connected. As edge switches are intended 2960G. In this case all traffic between desktops and servers pass through 32Gb StackWise bus. This “32Gb” is enough or not for previous examples?

This widget could not be displayed.