Re: Rate-limiting on Cat6K (off topic)

From: Daniel Shin (dshin@xxxxxxxxx)
Date: Tue Feb 27 2001 - 02:49:31 GMT-3


   
If it has a physical port associated with it such as your situation, it will
support it. I meant in a msfc that runs native IOS and has a vlan interface
associated with it.

> From dp595@optonline.net Mon Feb 26 12:15:04 2001
> Date: Mon, 26 Feb 2001 15:14:11 -0500
> From: Dan <dp595@optonline.net>
> Subject: Re: Rate-limiting on Cat6K (off topic)
> To: Daniel Shin <dshin@cisco.com>, ccielab@groupstudy.com
> MIME-version: 1.0
> X-MIMEOLE: Produced By Microsoft MimeOLE V5.50.4133.2400
> Content-transfer-encoding: 7BIT
> X-Priority: 3
> X-MSMail-priority: Normal
>
> This switch is at one of our datacenters and we have customers on the
> FastEthernet Ports.
> This switch is set up exactly like a router. Routing with OSPF and BGP. No
> Vlans defined (except for the default Vlan1 with Interface Vlan 1 shutdown),
> IP addresses on physical FastEthernet interfaces.
>
> Well I had set up the rate-limiting like this, but I wasn't sure:
>
> #sh ru int f8/3
> Building configuration...
>
> Current configuration:
> !
> interface FastEthernet8/3
> description 1 MB Customer Burstable to 10MB
> ip address x.x.x.x x.x.x.x
> rate-limit input 1000000 1125000 1125000 conform-action transmit
> exceed-action drop
> rate-limit output 1000000 1125000 1125000 conform-action transmit
> exceed-action drop
> no cdp enable
>
>
> It seemed to take, but I haven't plugged in yet and tested. Just going
> through the preliminary.
>
> #sh int f8/3 rate-limit
> FastEthernet8/3
> Input
> matches: all traffic
> params: 1000000 bps, 1125000 limit, 1125000 extended limit
> conformed 0 packets, 0 bytes; action: transmit
> exceeded 0 packets, 0 bytes; action: drop
> last packet: 2392521644ms ago, current burst: 0 bytes
> last cleared 01:42:55 ago, conformed 0 bps, exceeded 0 bps
> Output
> matches: all traffic
> params: 1000000 bps, 1125000 limit, 1125000 extended limit
> conformed 0 packets, 0 bytes; action: transmit
> exceeded 0 packets, 0 bytes; action: drop
> last packet: 2392521644ms ago, current burst: 0 bytes
> last cleared 01:42:39 ago, conformed 0 bps, exceeded 0 bps
>
>
> This Supervisor seems to support this:
>
> #show module
> Slot Ports Card Type Model
> Serial Number
> ---- ----- ----------------------------------------- --------------------- -
> ------------
> 1 2 Cat 6000 sup 1 Enhanced QoS (active) WS-X6K-SUP1A-2GE
> SAD04450AXT
> 8 48 48 port 10/100 mb RJ45 WS-X6348-RJ-45
> SAL04401A55
> 9 48 48 port 10/100 mb RJ45 WS-X6348-RJ-45 S
> AL04401ANL
>
>
> I am obviously not familiar with Cat6K.
>
> Dan Pontrelli
> Customer Installation Engineer - Verio NYC
> CCNP, MCSE, CNA
>
>
> ----- Original Message -----
> From: "Daniel Shin" <dshin@cisco.com>
> To: <dshin@cisco.com>; <ccielab@groupstudy.com>; <dp595@optonline.net>
> Sent: Monday, February 26, 2001 2:02 PM
> Subject: Re: Rate-limiting on Cat6K (off topic)
>
>
> > The first value is the actual rate limit. Second value is what you allow
> them
> > to burst up to. And the third one is if there is a credit from previous
> second,
> > then how much more would you allow them to burst. All the values added
> together
> > is your total car.
> >
> > However, going back to the first question, Catalysts do not support CAR.
> They
> > have vlan interfaces that are (literally virtual interface and there is no
> > physical port associated with them). When a train of packets arrive, the
> first
> > packet will go through the vlan interface to be routed and the rest of
> packets
> > get fastswitched at the wire rate on the switch. That's why the catalysts
> can
> > route packets so much faster than 7500 or 7200 due to that architecture.
> They
> > literally switch packets at a wire rate (the same switching as a layer 2).
> >
> > Downside is that the vlan never sees the full traffic going through. So
> CAR is
> > not possible on a catalyst because it will never reach the rate-limit to
> be
> > discarded.
> >
> >
> >
> >
> > > From dp595@optonline.net Mon Feb 26 10:40:48 2001
> > > Date: Mon, 26 Feb 2001 13:40:03 -0500
> > > From: Dan <dp595@optonline.net>
> > > Subject: Re: Rate-limiting on Cat6K (off topic)
> > > To: Daniel Shin <dshin@cisco.com>, ccielab@groupstudy.com
> > > MIME-version: 1.0
> > > X-MIMEOLE: Produced By Microsoft MimeOLE V5.50.4133.2400
> > > Content-transfer-encoding: 7BIT
> > > X-Priority: 3
> > > X-MSMail-priority: Normal
> > >
> > > 6509 Running IOS, and I think I had the burst rates wrong in my example
> > > below. I want 9MB of burst, therefore 1125000 bytes of burst (not
> 1250000).
> > > Am I using the correct logic here?
> > >
> > > Dan Pontrelli
> > > Customer Installation Engineer - Verio NYC
> > > CCNP, MCSE, CNA
> > >
> > > ----- Original Message -----
> > > From: "Daniel Shin" <dshin@cisco.com>
> > > To: <ccielab@groupstudy.com>; <dp595@optonline.net>
> > > Sent: Monday, February 26, 2001 1:12 PM
> > > Subject: Re: Rate-limiting on Cat6K (off topic)
> > >
> > >
> > > > Are you running hybrid or native IOS?
> > > >
> > > > Answers embedded...
> > > >
> > > >
> > > > > From nobody@groupstudy.com Mon Feb 26 09:25:22 2001
> > > > > Date: Mon, 26 Feb 2001 12:18:16 -0500
> > > > > From: Dan <dp595@optonline.net>
> > > > > Subject: Rate-limiting on Cat6K (off topic)
> > > > > To: ccielab@groupstudy.com
> > > > > X-MIMEOLE: Produced By Microsoft MimeOLE V5.50.4133.2400
> > > > > X-Priority: 3
> > > > > X-MSMail-priority: Normal
> > > > >
> > > >
> > > > CAR is not supported due to achitecture of catalysts.
> > > >
> > > > > I am trying to rate-limit a connection on a Cat6K FastEthernet
> port.
> > > > > I want 1MB bandwidth burstable to 10MB. I looked at the Cisco
> website
> > > and I
> > > > > think the following would do the trick:
> > > > >
> > > > > rate-limit input 1000000 1250000 1250000 conform-action transmit
> > > > > exceed-action drop
> > > > > rate-limit output 1000000 1250000 1250000 conform-action transmit
> > > > > exceed-action drop
> > > > >
> > > > > Is this correct? I'm still having trouble understanding why 2 burst
> > > rates
> > > > > (normal and maximum) need to be specified.
> > > > >
> > > >
> > > > If you don't, you will end up rate-limiting more traffic than
> intended.
> > > Think
> > > > about the traffic pattern. It bursts up and down - especially FTP.
> > > >
> > > > > Thanks in advance for any input.
> > > > >
> > > > >
> > > > > Dan Pontrelli
> > > > > Customer Installation Engineer - Verio NYC
> > > > > CCNP, MCSE, CNA
> > > > >
> > > > >



This archive was generated by hypermail 2.1.4 : Thu Jun 13 2002 - 10:29:07 GMT-3