Browse Prior Art Database

NIC NCP experiment (RFC0550) Disclosure Number: IPCOM000003654D
Original Publication Date: 1973-Aug-01
Included in the Prior Art Database: 2019-Feb-12
Document File: 2 page(s) / 3K

Publishing Venue

Internet Society Requests For Comment (RFCs)

Related People

L.P. Deutsch: AUTHOR

Related Documents

10.17487/RFC0550: DOI

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 56% of the total text.

Network Working Group L.P. Deutsch RFC # 550 PARC-MAXC NIC # 17796 August 24, 1973


For the past couple of weeks, the NIC NCP has been keeping statistics on total incoming messages, incoming host-host control opcodes, and size of outgoing messages. The results have been rather enlightening and, I think, should be carefully considered by future implementors of NCPs for servers. The statistics will be presented in a rather qualitative fashion, since they were reset each time the system came up, but they represent a total of about 100 hours of uptime, most of it during the working day.

The total numbers of incoming and outgoing messages were almost identical. There were about 5% more outgoing. There were slightly over half as many incoming control opcodes processed as incoming messages; on the assumption that no incoming control message had more than one opcode, slightly over half the incoming messages were control messages.

The Opcode statistics were somewhat variable. In all cases the ALL opcode accounted for the great majority, from a low of about 50% on weekends to a high of 98% on a busy weekday. Almost all of the remainder were NOPs. No other opcode ever accounted for more than 5%.

The output message statistics were taken as log2(message size): this included 1 word of buffer header, 1 word of IMP header, and l word of host header. As might be expected, 95% of all outgoing messages had l to 4 PDP-10 words (36-bit) of data. However, if one multiplies the count for each bucket by the average message site for the bucket, the result is that only 75% of all outgoing data was in the smallest message size: the remaining data was spread out fairly evenly between the other buckets.

I would draw the following conclusions from these statistics. First, half the messages on the network appear to be ALLs. This suggests th...