3850 VLAN interfaces limited to 1G?

[johnsmunoz]

I have a WS-C3850-48T.  I was doing a transferrer between two 10G interfaces and noticed my monitoring system was alerting me to high utilization.

 Looking close I saw that the VLAN shared between the two interfaces is only showing 1G.  A "show interface vlan 300" verified such.

eqx-c3750#show int vlan 300
Vlan300 is up, line protocol is up , Autostate Enabled
  Hardware is Ethernet SVI, address is 009e.1e63.0747 (bia 009e.1e63.0747)
  Internet address is 10.40.0.10/24
  MTU 1500 bytes, BW 1000000 Kbit/sec, DLY 10 usec,
     reliability 255/255, txload 226/255, rxload 1/255

Is it possible to get more throughput between TenGig interfaces?  I didn't see any speed options in the VLAN interface.  

[balaji.bandi]

as i per i know you can increase it, but default is 1GB  - what version of code - this will not effect the performance this only effects routing protocol.

[Joseph W. Doherty]

As already described by @balaji.bandi , your 10g interfaces are not limited by the SVI's default bandwidth of 1 gig.  Just use a bandwidth statement to increase to whatever value you wish to use as a base for percentage of utilization.

BTW, as in theory in a stack you could have same VLAN ports across the stack, and only limited by the stack ring bandwidth of 480 Gbps, you might use that as the bandwidth value (if you can set it that high).

[M02@rt37 · Accepted solution]

Hello @johnsmunoz,

When dealing with VLAN interfaces on Cisco switches, the default behavior is to display the bandwidth of the VLAN interface as 1 Gbps (`BW 1000000 Kbit/sec`), regardless of the actual physical interfaces' speeds (e.g., 10 Gbps).

To make the VLAN interface accurately reflect the speed of the underlying physical interfaces, you can use a "bandwidth statement" under the VLAN interface configuration. This statement is for informational purposes only and doesn't actually change the speed of the interface; it's more like a label. The Bandwidth statement in the VLAN interface configuration does impact how certain routing protocols and QoS treat the interface. While it doesn't physically change the speed of the interface, it influences how the switch and associated services perceive the bandwidth for various purposes.

In the context of a stacked switch environment, where multiple switches are stacked together, the stack's ring bandwidth (e.g., 480 Gbps for the Cisco Catalyst 3850 series) can be used as the bandwidth value. This reflects the combined capacity of the stack and gives a more accurate representation of the VLAN interface's capabilities in that specific setup.

[Joseph W. Doherty]

As M02@rt37 also mentions using a 480 Gbps value, as that's the StackWise-480 "bandwidth", I thought perhaps I should better describe how you might "correctly" configure the SVI's bandwidth.

Firstly, I believe Cisco quotes StackWise bandwidth's full duplex value.  So, for an interface bandwidth setting you would use half that value.  (Same as we use gig for gig Ethernet, not two gig.)

Second, you can not be exact because the SVI's is the aggregate of all that VLAN's port bandwidths, on the stack, possibly, but not necessarily, limited by StackWise bandwidth.

The second case might be difficult to understand.  Consider a stack where half ports in each stack member are in different VLANs.  Then suppose each switch port exchanges full throughput on just one other port, on SAME switch member, but other VLAN. I.e. StackWise not involved, but SVIs would show stack aggregate port bandwidths.

Change above such that each port's exchange mate is on another switch member, now StackWise might limit SVI bandwidth.

Since all the stack's same VLAN ports would be the actual maximum possible SVI bandwidth, that would be the "correct" value to use, but using the ring or fabric bandwidth is, perhaps, both simpler and "good enough".