From: Conte, Charles (Charles.Conte@nasdaq.com)
Date: Tue Feb 01 2005 - 23:59:20 GMT-3
Balaji,
Nice Link! Thanks!
Charles
-----Original Message-----
From: Balaji Siva [mailto:bsivasub@gmail.com]
Sent: Tuesday, February 01, 2005 9:02 PM
To: phase90
Cc: Conte, Charles; asadovnikov; ccielab@groupstudy.com
Subject: Re: Simple Design Question
Guess what ..that is already been done :)
http://www.cisco.com/warp/customer/cc/pd/si/casi/ca6000/tech/65acl_wp.pd
f
Look at ACL performance characteristics tests. That statement applies
to other switches as well.
My 2c to this thread is only about penalty on l2/l3 switching/routing..
:)
On Tue, 1 Feb 2005 20:54:24 -0500, phase90 <phase90@comcast.net> wrote:
> I understand Mr. Balaji. It looks as if this is something that would
need
> some lab
> benchmarking. Certainly I hope you would agree with me that a layer-2
config
> is
> much simpler. I guess you could static route it for 1 hop!
>
> Jerry
> ----- Original Message -----
> From: "Balaji Siva" <bsivasub@gmail.com>
> To: "phase90" <phase90@comcast.net>
> Cc: "Conte, Charles" <Charles.Conte@nasdaq.com>; "asadovnikov"
> <asadovnikov@comcast.net>; <ccielab@groupstudy.com>
>
> Sent: Tuesday, February 01, 2005 8:34 PM
> Subject: Re: Simple Design Question
>
> > ;) Note that my comment is in regards to the Catalyst switches
which
> > do everything in hardware level.
> >
> > Assuming the ACL is supported in hardware (i am not sure if turbo
acl
> > is supported in hw or not), L2 switching and L3 switching should be
> > same.
> >
> > So in other words if you have 2 L2 switching or 2 l3
> > switching/routing, it should be same for all practical purposes
AFAIK
> > :)
> >
> >
> >
> > On Tue, 1 Feb 2005 20:22:15 -0500, phase90 <phase90@comcast.net>
wrote:
> > > So if I read this correctly, a packet going through 2 routing hops
and a
> > > 2000 line [ turbo ] ACL
> > > has the same latency as a packet going 0 hops via connected Vlan
> interface.
> > > I think you've been
> > > talking to the Cisco marketing group too long!
> > >
> > >
> > > phase90
> > > ----- Original Message -----
> > > From: "Balaji Siva" <bsivasub@gmail.com>
> > > To: "Conte, Charles" <Charles.Conte@nasdaq.com>
> > >
> > > Cc: "phase90" <phase90@comcast.net>; "asadovnikov"
> > > <asadovnikov@comcast.net>; <ccielab@groupstudy.com>
> > > Sent: Tuesday, February 01, 2005 7:38 PM
> > > Subject: Re: Simple Design Question
> > >
> > > > Yes that is correct. There is no speed penalty for L2 or L3
switching.
> > > > For example on cat4k, it is all done in hw asic and if the
packet is
> > > > not routed, that function is "no opearation".. So whether you
turn on
> > > > routing/acl/qos, it all is same.
> > > >
> > > >
> > > >
> > > >
> > > > On Tue, 1 Feb 2005 19:34:32 -0500, Conte, Charles
> > > > <Charles.Conte@nasdaq.com> wrote:
> > > > > Hello Phase,
> > > > >
> > > > > My reason would be to avoid spanning-tree. With Layer
3
> > > > > switching there is practically no difference in latency.
> Spanning-tree
> > > > > is a lot harder to troubleshoot in situations of a loop. I
like the
> > > > > document below on how it talks about some aspects of spanning
tree.
> I
> > > > > guess everything has the "it depends" attached to it. :)
> > > > >
> > > > >
>
http://www.cisco.com/en/US/tech/tk389/tk621/technologies_tech_note09186a
> > > > > 00800951ac.shtml
> > > > >
> > > > > CISCO DOCUMENTATION:
> > > > > High-end Cisco Layer 3 switches are now able to perform this
second
> > > > > function, at the same speed as the Layer 2 switching function.
There
> is
> > > > > no speed penalty in introducing a routing hop and creating an
> additional
> > > > > segmentation of the network.
> > > > >
> > > > >
> > > > > -----Original Message-----
> > > > > From: phase90 [mailto:phase90@comcast.net]
> > > > > Sent: Tuesday, February 01, 2005 7:17 PM
> > > > > To: asadovnikov; Conte, Charles; ccielab@groupstudy.com
> > > > > Subject: Re: Simple Design Question
> > > > >
> > > > > Yes but what if your access switch / router is one hop from
your
> core,
> > > > > why
> > > > > would you route that hop and have the additional latency in
the
> routing
> > > > > process?
> > > > >
> > > > > Jerry
> > > > > ----- Original Message -----
> > > > > From: "asadovnikov" <asadovnikov@comcast.net>
> > > > > To: "'Conte, Charles'" <Charles.Conte@nasdaq.com>;
> > > > > <ccielab@groupstudy.com>
> > > > > Sent: Tuesday, February 01, 2005 12:57 AM
> > > > > Subject: RE: Simple Design Question
> > > > >
> > > > > > I like the approach. If access switches are L3 capable you
should
> run
> > > > > them
> > > > > > as routers not switches. Although there are always corner
cases
> when
> > > > > L2
> > > > > may
> > > > > > be better option, I strongly agree that benefits of avoiding
L2
> > > > > generally
> > > > > > greater then any potential downside.
> > > > > >
> > > > > > Best Regards,
> > > > > > Alexei
> > > > > >
> > > > > >
> > > > > > -----Original Message-----
> > > > > > From: nobody@groupstudy.com [mailto:nobody@groupstudy.com]
On
> Behalf
> > > > > Of
> > > > > > Conte, Charles
> > > > > > Sent: Monday, January 31, 2005 7:29 PM
> > > > > > To: ccielab@groupstudy.com
> > > > > > Subject: OT:Simple Design Question
> > > > > >
> > > > > > Hello,
> > > > > >
> > > > > >
> > > > > >
> > > > > > If MSFC's are available at the access-layer, can
> anybody
> > > > > > tell me why we wouldn't run L3 to the access layer if the
primary
> and
> > > > > > secondary access switches are available in convenient
locations?
> Also
> > > > > > for the attached gifs can anybody provide any opinions on
why one
> > > > > > wouldn't extend L3 to the access instead of having L2 only
Access
> > > > > > switches [Example 1 L3] V.S. [Example 2 L2]? I like
avoiding L2
> in
> > > > > any
> > > > > > situations that I can. I can understand if the requirement
is to
> have
> > > > > > the vlan available at every switch to go with example 2, but
if
> not it
> > > > > > wouldn't make sense to extend L2 everywhere. Any opinions
> > > > > appreciated!
> > > > > > Thanks!
> > > > > >
> > > > > >
> > > > > >
> > > > > > Charles
> > > > > >
> > > > > > [GroupStudy removed an attachment of type image/gif which
had a
> name
> > > > > of
> > > > > > example_gif_2.gif]
> > > > > >
> > > > > > [GroupStudy removed an attachment of type image/gif which
had a
> name
> > > > > of
> > > > > > example_gif_1.gif]
> > > > > >
> > > > > >
> > > > >
>
This archive was generated by hypermail 2.1.4 : Thu Mar 03 2005 - 08:51:16 GMT-3