Very Computer

Has anyone ever gotten good results using slip to an hp box (hp calls
this ppl)?

What type of line are you using? 1200, 9600? 19.2? Is there other
traffic on the line at the time? Are pings lost? Are you sending
larger than default pings?

The response time seems bad, what is the throughput like? (For
measuring throughput, and/or response time, people might want to look
at netperf, available via anonymous FTP from
col.hp.com:dist/networking/benchmarks grab version 1.7, the 1.8 that
is out there doesn't have the updated docs...oops)

I can remember playing with ping on MPE with X.25, and with a suitably
slow link, I could back-up the line with pings and watch the response
time grow...

rick jones

    Rick> What type of line are you using? 1200, 9600? 19.2? Is there
    Rick> other traffic on the line at the time? Are pings lost? Are
    Rick> you sending larger than default pings?

Thanks for the reply.  The hp is set at 38.4 with rts/cts using a
worldblazer.  I tried the linux box (also using a worldblazer) set to
19.2 and to 9600.  Pings are not lost.  I did not try an objective
measure like netperf yet, but the link is useless.  Telnet to the hp
server is so slow it's useless.

To vaguely summarize:  If you run PEP mode, you get poor performance over
SLIP links.  It is something like the packets SLIP is sending are causing
the modem to wait for timeouts to bundle them up in PEP packets to
send over the phone line.  I seem to recall that *not* using VJ header
compression actually makes things faster.

I recommend that anybody using a worldblazer get this document.  Till then,
you might just try disabling PEP mode and see if it gets any better.

Good luck,

Mark.

Quote:>Telebit has a README about how their modems behave in the face of SLIP
>traffic-- you can get it from ftp.uu.net (in vendor/telebit or something
>ike that).

---

--------------------------
Derek J. Pietro
Picker International