[c-nsp] OC3 Throughput

[Robert E. Seastrom]

Clayton Zekelman <clayton at mnsi.net> writes:

> An NPE-G1 easily handles 2 full ATM-OC3's worth of L2TP tunnels.
>
> We have several of these in production, and typically hit around 60%
> CPU utilization.

My experience with both NPE-G1s and 7301s does not parallel this.
Could be an artifact of the load we were running, the fact that we
were running on 802.11 vlans on gigabit ethernet rather than ATM, or
some kind of local configuration requirement that I haven't thought of
today.  Wasn't running unicast RPF on the NPEs, preferring to run it
on the edge-facing interfaces in the GSRs.  Was running dynamically
loaded outbound SMTP filters via radius attribute 242.  Yes, I did
have CEF turned on.  :-)

My experience is that once you were in the 120-130 Mbit range (sum of
in and out), you were pretty much at the 80+% range and should be
looking at a new router to add to the mix.  Of course, it probably
depends even more on packets per second than it does on data moved.

My customers were some filesharin' fools.  In the early morning our
outbound exceeded the inbound by about 20%.  That could have figured
into our situation as well.

Obviously caution is required for the budgetary numbers; take Cisco's
and your colleagues' numbers with a grain of salt, hedge the numbers
in your model, and come out looking like a hero.  :-)

                                        ---Rob

>
> ----- Original Message ---------------
>
> Subject: Re: [c-nsp] OC3 Throughput
>    From: "Robert E. Seastrom" <rs at seastrom.com>
>    Date: Mon, 26 Nov 2007 07:48:40 -0500
>      To: Aaron <dudepron at gmail.com>
>      Cc: cisco-nsp at puck.nether.net
>
>>
>>Of course, the *real* answer which everyone seems to be overlooking is
>>that you're terminating PPPoE-over-L2TP per Paul's original mail.  The
>>encap/deencap is the limiting factor, and you're gonna pummel any
>>known NPE up to and including the NPE-G1 before you hit the link speed
>>limit with ATM or POS (you're running an ISP; you *will* have traffic
>>going both ways, and substantial upstream in todays p2p world).
>>
>>What processor do you have in there?
>>
>>                                        ---Rob
>>
>>Aaron <dudepron at gmail.com> writes:
>>
>>> Of course you didn't specify ATM originally either. POS would probably
>>> give you a few more since you don't have the ATM overhead.
>>>
>>> On Nov 17, 2007 3:09 PM, Garry <gkg at gmx.de> wrote:
>>>> Paul Stewart wrote:
>>>> > A few people hit me offline stating that it's not nice to hold back the
>>>> > answers and that they were curious... right you are - sorry about that...;)
>>>> >
>>>> > The general answer has been 110-120Mb/s considering ATM overheads etc. on a
>>>> > 5 minute graph average - beyond that you are really pushing your luck.  We
>>>> > have one such circuit today that is hitting 110 regularily and sometimes
>>>> > hits 120 for short peaks - my concerns are now staring at me on a graph....
>>>>
>>>> In general (at least speaking from an ISP's point of view) when
>>>> utilizing a link beyond 90% peak or beyond 75% average, it's time to
>>>> upgrade ... otherwise it's just a question of time when the link
>>>> capacity is causing quality degradation for the users ... (unless the
>>>> peaks are caused by batch-only traffic).
>>>>
>>>> Take the usage graph to your boss and tell him you need another link or
>>>> an OC12 upgrade ;)
>>>>
>>>> -garry
>>>>
>>>> _______________________________________________
>>>> cisco-nsp mailing list  cisco-nsp at puck.nether.net
>>>> https://puck.nether.net/mailman/listinfo/cisco-nsp
>>>> archive at http://puck.nether.net/pipermail/cisco-nsp/
>>>>
>>> _______________________________________________
>>> cisco-nsp mailing list  cisco-nsp at puck.nether.net
>>> https://puck.nether.net/mailman/listinfo/cisco-nsp
>>> archive at http://puck.nether.net/pipermail/cisco-nsp/
>>_______________________________________________
>>cisco-nsp mailing list  cisco-nsp at puck.nether.net
>>https://puck.nether.net/mailman/listinfo/cisco-nsp
>>archive at http://puck.nether.net/pipermail/cisco-nsp/