Re: CHGTCPA question

[pytel@xxxxxxxxxx]

>    1)  TCP keep alive       120
>    2)  TCP urgent pointer        *BSD
>    3)  TCP receive buffer size   8192
>    4)  TCP send buffer size 8192
>    5)  UDP checksum         *YES
>    6)  IP datagram forwarding    *YES
>    7)  IP source routing         *YES
>    8)  IP reassembly time-out    120
>    9)  IP time to live      64
>    10) ARP cache timeout         5
>    11) Log protocol errors  *YES


>    I think 1 & 2 can be left alone.

Yes

>    3 & 4 seems a little high.
>    I suspect that they are seeing noticeable degradation when there are
line errors.

It depends. By 3 and 4 you control transmission window size.
If network (or NIC) is congested, then decrease these - this will improve
performance by reducing retransmissions.
If congestion is not an issue, then increase these - this will improve
performance by reducing turnarounds.

>    Can 5 - 7 safely be changed to *NO?  Would it help performance?

You can change 5) to *NO, if you are confident in network quality. But
resulting performance improvement will be minimal.
6) and 7) do not influence performance.

>    What are 8 - 10?

If you have *very* poor network quality, when lots of packets are regularly
lost, you may want to decrease 8) - this will speed up retransmission
detection.
But if you have good but slow network, don't do this - you will increase
unnecessary retransmissions.

Defaults for 9) and 10) are normally OK.

>    11 is OK.  Actually, does this need to be checked in SST?

Yes. Protocol errors, if any, will be logged in Product Activity Log.


Alexei Pytel


+---
| This is the Midrange System Mailing List!
| To submit a new message, send your mail to MIDRANGE-L@midrange.com.
| To subscribe to this list send email to MIDRANGE-L-SUB@midrange.com.
| To unsubscribe from this list send email to MIDRANGE-L-UNSUB@midrange.com.
| Questions should be directed to the list owner/operator: david@midrange.com
+---