On Fri, 13 Oct 2000, Brian W. wrote:
> I disagree, I know of plenty 2 and 3t1 customers of a former employer that
> pushed aggregate bandwidth in excess of the 1.5megabits you claim. In
Aggregate bandwidth yes........but no single user on a single session,
like a downlod for example, would ever see more than 1.5Mbps with
per-destination or per-session load-balancing. What I am saying is that
without per-packet, no "single" session or user can source more than
1.5Mbps.
> those cases, I occasionally got calls from managerial types that wanted
> to see symmetric graphs. If both circuits are being used, thats good
> enough for me. Your argument only makes sense if for an individual
> established connection, bandwidth greater than that provided by 1 circuit
> is not enough.
this isn't true. per-packet is about alot more than pretty graphs. Its
the difference in maximum speed. 3 T1's in per-session load balancing can
push a total of 1.5MBps x 3, but still only peak at 1.5Mbps for any
session. I admit its confusing, but take my word for it, per-packet is
the only way to break the single circuit barrier. Otherwise you have a
"flow" (route-cache flow) or a destination (cef per-destination) which is
cached over a SINGLE t1 circuit........and it will ONLY use that circuit
for the entire flow or destination.
Brian
>
> Brian
>
> On Fri, 13 Oct 2000, Brian wrote:
>
> > On Fri, 13 Oct 2000, Brian W. wrote:
> >
> > > imho, balancing on a per packet basis is for managerial types that like to
> > > see symmetrical graphs. I beleive you have to disable caching on all the
> > > affected interfaces to get it, a basically bad idea imho.
> >
> > not hardly. If you have just 2 PtP T1's to a remote site for example,
> > with clients that can source lots of data, you will be wasting alot of
> > bandwidth without per packet. You could have one T1 with a session that
> > is tapping out, and the other t1 is at 50% and nothing can be done about
> > it.
> >
> > also, if you have 4 t1's with perpacket on them, you effectivly have 6Mbps
> > of bandwidth to the site. With 4 t1's and "per session" or "per
> > destination" load balacing, you can only hope to source 1.5Mbps at most.
> >
> > Per-packet is far from a gimmick, hence the hardship people are willing to
> > go thru for CEF
> >
> > >
> > > Brian
> > >
> > > On Fri, 13 Oct 2000, [iso-8859-1] Phil Barker wrote:
> > >
> > > > Hi groupies,
> > > > Can anyone enlighten me on the process by which
> > > > per-packet load-balancing re-assembles the packets at
> > > > the remote router. Which sequence number is used e.g
> > > > is it the TCP sequence number.
> > > >
> > > > Regards,
> > > >
> > > > Phil.
> > > >
> > > >
> > > >
> > > > ____________________________________________________________
> > > > Do You Yahoo!?
> > > > Get your free @yahoo.co.uk address at http://mail.yahoo.co.uk
> > > > or your free @yahoo.ie address at http://mail.yahoo.ie
> > > >
> > > > _________________________________
> > > > FAQ, list archives, and subscription info:
>http://www.groupstudy.com/list/cisco.html
> > > > Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
> > > >
> > >
> > > _________________________________
> > > FAQ, list archives, and subscription info:
>http://www.groupstudy.com/list/cisco.html
> > > Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
> > >
> >
> > -----------------------------------------------
> > Brian Feeny, CCNP, CCDP [EMAIL PROTECTED]
> > Network Administrator
> > ShreveNet Inc. (ASN 11881)
> >
>
-----------------------------------------------
Brian Feeny, CCNP, CCDP [EMAIL PROTECTED]
Network Administrator
ShreveNet Inc. (ASN 11881)
_________________________________
FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html
Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
