[c-nsp] Packet Latency Calculator

[Bruce Pinsky]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Eric Helm wrote:
| Matthew Crocker wrote:
|
|>Hello,
|>
|>   Every once and a while I have a customer complain about latency
|>through our network.  Normally it is on the initial hop (DSL).  Does
|>anyone know of a calculator I can use to figure the theoretical
|>minimum latency for a circuit?
|>
|>For example:
|>
|>Customer traceroutes through to their gaming server (15 hops away).
|>They see 30ms ping times on the first hop (DSL Line).  The DSL line
|>is IP over PPP over Ethernet over ATM over DSL.  With all the extra
|>encapsulation and the ATM cell tax how big is the new packet?  How
|>long would it take to send/recieve this packet over a 768kbps DSL link?
|>
|>Anyone know a good place to look?
|>
|>-Matt
|>
|>_______________________________________________
|>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/
|>
|
|
| Matt,
| This cisco link has a nice table of serialization delay values:
|
http://www.cisco.com/warp/public/788/voip/delay-details.html#serializationdelay
|
| And here is discussion of Protocol Overhead for Ethernet and ATM:
| http://sd.wareonearth.com/~phil/net/overhead/
|
| I belive PPPoE adds 8 bytes of overhead.
|

The actual PPPoE header is only 6 bytes and the PPP protocol ID accounts
for 2, but there are other contributors that can result in up to 20 bytes
of overhead:

- - PPPoE header (6 bytes)
- - largest possible outer PPP header (4 bytes)
- - largest possible Multilink PPP header (4 bytes)
- - largest possible PPP header for compression and encryption (4 bytes)
- - PPP header that identifies the actual packet being sent (2 bytes)


In addition to serialization delay, there is also transit delay which is
governed by both distance and the speed of electrons, depending on the
medium being used for transmission.  Through fiber, light travels at about
200,000 km/s = 125,000 mi/s. So 5000 miles worth of fiber (which could be
the Atlantic for example) is ~40 ms (and double for RTT).  I also
understand that different DSL technologies and modes can affect
tranmission, for example ADSL interleaved vs non-interleaved.

Interesting spreadsheet I found at
http://www.monkeynoodle.org/comp/tools/Data_Replication_Time_Calculator
that includes a number of factors for determining how long it will take to
replicate a large amount of data from one site to another over various
speed links.  It allows you to change the various factors and see the
result.  While not exactly what you may be looking for, it could serve as a
basis for modification.

- --
=========
bep

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.0 (MingW32)

iD8DBQFCeRK0E1XcgMgrtyYRAqXZAKChd05bR91IB4R51QC7JtLacGHjFgCfbFtI
5RDoBdv87FlZwRj0VSEnI44=
=pUFM
-----END PGP SIGNATURE-----