RFC101 Note on the network working group meeting of Urbana in Illinois on February 17, 1971

xiaoxiao2021-03-06  151

Organization: China Interactive Publishing Network (http://www.china-pub.com/)

RFC Document Chinese Translation Program (http://www.china-pub.com/compters/emook/aboutemook.htm)

E-mail: Ouyang@china-pub.com

Translator: Shao Yi (EPL shaoyi@163.net)

Translation time: 2001-10-11

Copyright: This Chinese translation copyright belongs to China Interactive Publishing Network. Can be used for non-commercial use free reprint, but must

Keep the translation and copyright information of this document.

Network Working Group Richard W. Watson

Request for Comments: 101 SRI-ARC

NIC: 5762 February 23, 1971

February 17, 1971 Illinois Urbana

Network Working Group Conference Comment

(RFC0101 --- Notes on the network working group meetings)

table of Contents

On Wednesday, February 17

On Thursday morning, February 18

Thursday afternoon, February 18

Thursday night, February 18

Friday morning, February 19

On Wednesday, February 17

Mike Sher first welcomes everyone to come to Urbana, then briefly point out "Illiac Fourth" plan is expected

It started working in the summer. The "Illiac Fourth" program is divided into two subprints: one of them focuses on the basic system

The unnecessary hardware and software is related to the application. Their IMP has not been connected to its PDP-11.

Steve Crocker requires a meeting to discuss relevant issues, which are given in subsequent parts of this article.

Peggy Karp from Mitre has been committed to the summary of the old RFC documentation. She gave a given

About 30 entries list, and summarized their current state. Her work is expected to be around February

carry out. (See RFC # 100, NIC5761.) Everyone recommends that some people should be banned by those old RFC documents.

A RFC file. People also recommend that the network information center (NIC) is mentioned in terms of tissue hard copy materials.

Help.

Next is a brief discussion on the experience of network usage. John Melvin (SRI-ARC) puts SRI

Experience is summarized as uses it to help SRI from one XDS 940 to PDP-10. In 1970

From April to May, it is clear that SRI is developing towards PDP-10 to make themselves and reliability.

Exercise yourself as a task of the network information center. They have some experiences connected to UTAH before, therefore trying

It is also logically using UTAH 10 to assist its transition.

The use of the network begins in June. Since SRI uses a lot of senior languages, the primary task is

The translation of the information tree for the compiler. Generate source code for PDP-10 on 940, then binary text

Part transfer to UTAH debugging, and change source code in a possible place, add patch, and generate new

Source code and binary files.

When running information tree, a new high-level language for online system (NLS) programming (called L-10)

It is also achieved in the same way. When the L-10 is running, the NLS core device has been rewritten and tuned.

test. During the transition, NLS is completely reorganized.

In the end of SRI and UTAH, people have written a control program that allows three users to connect to UTAH. This control program is run as a user process, allowing character input and file transfer. This plan is running

Strain, and completed a lot of useful work in July to December. During this time, there are some people every day.

For 4 to 5 hours. Voice links can appear when trying to determine the problem and the re-combined work

Use when you are wrong. There is sometimes no mistakes last for two weeks. SRI has a process of running as a T / S process

IMP diagnostic interface.

In general, the response is processed at the SRI side. UTAH terminal is DDT. Retrieval character four seconds

It is not uncommon, and there is still a case where there is still a delay of 8 or 10 seconds in some places. Due to the implementation of each terminal

Revenue multi-process, and each process is predetermined, thereby generating these delays. Utah at two o'clock in the afternoon

The load is heavy, and SRI likes to run in the evening and weekends.

When SRI PDP-10 appears in December, the use of networks is slowed down.

Users will like a more constant reaction time, not a big change. This way, despite this time

Slowly, the user's work habits can also adapt to it.

Gerry Cole reports the results of some measures employed during the SRI-UTAH work. SRI also offers

Some measures help UCLA to explain the data obtained. Gerry wrote an article to summarize these statistics from SDC

data.

GERRY requests when people want to use the network to inform him to make him collect statistics. UCLA

A program for scanning the network will eventually use. But if people notify him when they will use the network,

The work will become easier to measure the meaning of meaningful information and the knowledge from a certain type of application.

Bob Kahn pointed out that BBN is very interested in the statistics of global information flow. They want to determine the network through this

Whether the configuration is appropriate. Gerry said that UCLA is interested in statistics on network model research. For this function,

Remote control uses methods such as IMP design features.

Jim White from UCSB said that UCSB and Rand have begun in RAND climate research work.

Test network. All the dayd days of the UCSB NCP have been activated during the three to four weeks. A number NIC5480

Describe the documentation of the application can be found in the NIC article. UCSB also uses their NCP to communicate local processes

Letter experiment. RAND is in remote work portfolio using UCSB 360-75. They use UCSB to check them

NCP. Now, UCSB runs its NCP in the normal use time. They found their hardware

Some errors in its IMP interface. UCSB and RAND's software seem well. Clicked the class

Type work is just a test task for some source statements. UCSB's NCP is approximately 39K, running on a 60K byte

Partition. Users access it through assembly language, fortran or pl / i calls.

Steve Crocker turns back to discuss the agenda of the meeting and the long-term organizational structure of NWG. Steve feels

Because of the opening of the meeting, general discussions and education are very beneficial, but the scale is too large

A detailed description of a wide variety of topics can be prepared, so actions to establish a working committee for different topics.

The following is the subject of demand work:

1. Graphics

2. Data transmission language

3. Host ---- Host Protocol (Long-term Research)

4. Host ---- Host Protocol (short-term maintenance and modification)

5. Calculate

6. Log agreement

7. Printer connection protocol

8. Documentation

9. Data management

About the first theme, MIT's Al Vezza is organized a plan to be held from April 25 to 27.

You can accommodate 31 people's graphics NWG meetings. People who want to participate in the meeting need to prepare for his unit

Work report. Al gives three types of issues:

1) Two hosts, each have calculation and graphics devices, and want to use some of the other special devices 2) A host with a graphics device without having a data processing device, I hope to use the second host

Calculate ability

3) A junction with a graphical terminal, but no graphics processing and computing power, wants to be from other nodes

Get graphics and computing power

About the second theme, John Heafner from RAND pointed out that Rand wishes to provide RFC83

The described data reconfiguration service is described. There is still more discussion about this topic.

About the third topic, A. N. Habermann from CMU has set a working group research.

Host ---- Host Protocol. Since the end of March, they have been preparing an article about their views. This work

The group consists of the following person:

A. N. Habermann, CMU

G. B. Hansen, CMU

W. WULF, CMU

R. Chen, CMU

R. Kalin, Lincoln Lab

Welcome to this topic.

Regarding the fourth topic, people have established a group to evaluate the current protocol and make the necessary modifications.

This working group will adopt a conservative working method, only to solve the necessary changes to a known issue, and change the so-called aesthetics

(Esthetic Changes) is reserved.

Discussion on other issues will be placed in the final.

Two people who are interested in networks and listen to the meeting are briefly spoke.

C. D. (Terry) of the Canadian government's computer exchange specializes in the goal of its working group

Overview. These goals include:

1) Establish a plan to connect a variety of Canadian computers to establish a network

2) Develop such a network for Canada's demand

3) Observe the distribution of this network in Canada

4) Prevent the control of Canadian computing technology, depending on the foreign resources

5) Observe the key computer facilities owned by Canada

Next, Doug Mckay from IBM briefly describes a network plan that IBM two years ago.

The base network has been completed and some users have. The network is frequently used for file round trip transmission when the program is upgraded.

IBM tries to view the network as a multiprocessor machine. They try to process all of the heterogeneous systems in IBM, such as

360's, 370's, CP'67, 91, 44, and a NYU CDC 6600.

There is another item using a 91 connected to the TSS system from the remote working entry. IBM uses one

Central machine distributed in control and information flows to master the concentrated control perspective. They are not very full of this program.

The intention, and is developing more concentrated programs similar to the ARPA network. At present, IBM has approximately 14 people.

In the project.

Thursday morning, February 18

On Thursday morning, people began reporting their working conditions in different locations. Alex from BBN

Mckenzie prepared a status table later, filled out by representatives of the meeting on Thursday evening. BBN with

The NIC plans to prepare a program that retains information at all locations and keeps an updated program.

situation

BBN, Tenex Program: The final phase of Tenex and NCP cooperation. Try to build a with UTAH

Connections, but found some errors. NCP treats the network as a file, one with other types of files

the way. NCP introduces a traditional interface. They hope to implement NCP in the TENEX system at SRI at the end of the month.

Merge in the middle.

BBN, network group: report three areas of their work:

1. Improve existing network

2. Work based on IMP316 version and work as a terminal interface processor (TIMP)

3. Calculate

Currently, there are 15 IMPs connected to the network. A new, software system that has made smaller changes will be completed in March.

TIMP uses 316 system. Although there is a hardware design, they still need to define the software environment. One

TIMP can solve until 64 different speed synchronous and asynchronous terminals. The first machine will arrive in September

Mitre.

BBN emphasizes three products: a 516Imp, a 316Imp and a 316timp. 316Imp

516Imp is inexpensive and can be connected to a host. BBN is currently not intended to convert 316IMP to

516Imp. Both are plug-in plugged.

SDC: It is still in the commissioning phase of their NCP, which is planned for four to six weeks. Maybe eight weeks later, their

T / s is available for networks. Their T / S is a 360/65 machine that runs the ADEPT system.

Case: IMP has been connected for nearly one month, but there is no NCP. They plan to pick

NCP used by Harvard University. Case has a PDP-10/50 system, which is expected to be completed within two to three months.

Harvard University: Harvard University has a PDP-1 and a PDP-10 connected to IMP. PDP-10 NCP

In the final debug stage. PDP-1 is used to refresh the display, and PDP-10 is used for language research and for students.

It is scheduled to be completed within one to two months.

SRI-ARC: SRI is at the final stage of the XDS 940 to PDP-10 conversion. They plan to use BBN

Tenex NCP. The program is expected to be completed in three to four weeks.

Mit dynamic modeling ---- PDP-10: They want an NCP to be run in March.

MIT's Multics: They have been connected to IMP, then four weeks, their NCP can enter

Enter the final debug stage. Since Multics is a server, they do not have unlimited access capabilities, and

And must be exited in the rest time. They plan to provide normal services for the network within three to four weeks.

Utah: PDP-10/50 is extremely likely to eventually run Tenex. Their NCP has been in a high-level language

Write and debug with BBN. They have completed their connectivity and landing. A debugging version will be in the month

Incidentally occurred.

Lincoln Lab, TX-2: They are testing IMP interfaces to discover issues in Linken system hardware. Objective

In addition to the message identification error, there is no data error. The NCP with a log system is expected to be four

The month is completed on the 15th. It is alleged that they will be based on experimental testing, open up without exposure to network transmission.

Its IMP. BBN said there is a method to be in response to itself without disclosing a network.

Lincoln Lab, 360/97: Running CP / CMS. The IMP interface has been completed last month. NCP and log

The server is working. They plan to establish NCP in April as a normal service. Experiment with them

The request can be implemented soon.

UCSB: It has already had its NCP since October last year. NCP runs as a separate batch task.

They plan to provide services to their online systems. They are planning to run throughout the day. At the same time

Some previously mentioned interface issues.

Rand: 360/65. Their NCP is a user process and can reside. This process takes 8K

Bytes and no logs need.

UCLA, SIGMA-7: Their NCP is in the final debug phase. They plan to build on March 1

NCP, log service, and print connector.

American Computer Co., Ltd. (CCA): It just started to create a node for 10 to 12-bit laser storage. in

The front end, they use PDP-10. They are developing languages ​​for data operations. The memory will also be

B-6500-IlliC IV is connected. They are considering compressing data as part of their language to reduce usage network

The problem when 50 kilocated cables. Its current work focuses on security and confidentiality measures. The initial focus is the document

enjoy. Its installation has been planned in 1972. The following plans to say that there is no representative at the meeting, and Steve Crocker reported their situation.

CMU: PDP-10/50: IMP has been connected, planned to use Harvard University NCP.

SRI-AI program: PDP10. Plan to run Tenex.

Stanford AI Plan: Not yet connected, plan this summer.

The above is a review of the current situation. Steve crokeer Next, the old NWG mailing list will no longer make

For use, a list of maintained by NIC (5731), or NIC can also be used by the site agent

The way information is sent to publish. If your site agent or contact person has changed, please notify

NIC.

Host ---- Host Log Protocol Discussion: Tom Skinner from Multics first speaking. He pointed out that they

At least one excessive protocol is required to start the network. They submitted the RFC98 document

(NIC5744), including their views on Wednesday. SRI-ARC also has a similar file, RFC97

Document (NIC5740) and submitted on Wednesday night. Mutics gives RFC80 files (NIC5608)

Revised log agreement.

Some relative advantages compared to the RFC66 document and the RFC80 files are also given. 80

The file agreement has some potential problems, because "work must be established in the first time" this fake

The reason is set.

The result of the discussion is the basis for the revision of the RFC66 file "establishment of the connection"

Accept your log agreement.

It seems that it seems necessary to establish an official document to correct the log description given.

Tom also recommends the first communication with the log service accounts for the 7-bit ASCII code in 8-bit fields. Whether people will be the eighth

Some discussions were carried out for 0 or 1, and finally, the eighth bit should be set to 0.

Next, Steve lists some of the problems existing in the host ---- Host protocol.

1) Response

2) Message type

3) Interrupt

4) Tags and fill

5) Comparison of half-duplex and full-duplex communication during the establishment of the interface

Considering the following options:

a) idle

b) divide the head and data into two information

c) Set the message to a 72-bit length multiple

Taking into account the interruption (INR), there is a synchronization problem associated with the message delivery. That is, a message can be

Send and issue an interrupt. Interrupts may arrive before the message, or may arrive in the middle of the message. Therefore need

It is necessary to indicate where to be interrupted in the data stream.

The meeting appointed a small group to consider the above host ---- host problem. In short, they will give a pair of main

The RFC documentation revised by the machine--host protocol, then collects opinions and suggestions, and submits an official revision.

People with suggestions or opinions should contact the Committee. The committee is also connected to the site. The committee is composed of the following person:

S. CROCKER, UCLA (Chair)

R. Tomlinson, BBN

T. BarkAlow, Lincon Lab

G. Grossman, University of Illinois

J. White, UCSB

R. Bressler, MIT, MAC program

Next, the discussion of the meeting returns to the access problem of network printing. This issue is in RFC97 (NIC5740)

It is mentioned in it. Some of them are:

a) character set

b) Row tail logo

c) interrupt

d) message format

e) Half duplex, full duplex

These issues are submitted to a print connection protocol committee to seek solutions. This committee consists of the following person:

Tom O'Sullivan, Raytheon (Chair)

Ed Meyer, MIT-Mac

John Melvin, SRI-ARC

Bob Long, SDC

Bob Metcalfe, Harvard

WIL Crowther, BBN

This committee will provide an excessive agreement within a week and provide a long-term association in the next few weeks.

.

Thursday afternoon, February 18

On Thursday afternoon, IllinoIS University reported on the fourth plan of Illiac, in addition to the Plato program

Demo. The preliminary test of the line to transmit to the ILLIAC fourth dispatcher last November did not find

Time problem. The hardware and software of the ILLIAC fourth system will eventually be established. The system will be placed in California

NASA Ames Research Center. The connection from the University of Illinois to the network will consist of a CRTS memory,

2400 baud character CRTS, and PDP-11 implementation of additional input devices. It will install a gould at the same time

Clevie printer and DEC tape drive and small disc machine. The University's B6500 will also be connected to the network.

Thursday night, February 18

The initial issue is the discussion on the status and planning of the network information center. Dick Watson from SRI

Retrospects the offline system that is currently consisting of site proxy and network contact personnel. The function of the site agent is auxiliary NIC

The service is used. The role of network contact person is to ask the staff asked by the person from other sites.

The contact point of the question, in addition, he also monitors whether the appropriate person can consult the documents and information accepted by the site. Such as

If the network should create a community feel, then people need to know what they are doing, and from different sites

Conduct. Therefore, people encourage people to submit reports, memo, notes, and some interested in NIC.

Record of the discussion of the problem. From the official report to the informal manuscript can be submitted to the NIC. In order to encourage people to submit the most

At the beginning of the idea and opinion, NIC gives a question, namely, whether there should be a title to the documents to be titled.

Level of formal and informal communications.

It seems that this arrangement is not necessary. Secondly, issues are privacy and security. some

People think that if letter or conversation record is included in the NIC article, some of them will contain a

People privacy. People ask if NIC will check all aspects of such communication prior to inclusion. Dick

It is considered that since there is a NIC resource existence, then all aspects involved are given to the NIC.

Maybe it will be better.

The initial online service provided by NIC is a visit to a SRI-ARC online system (NLS) typing version.

ask. Provide people with message service, access NIC's directory, and possible site status documents, and network

Staff, etc. The service will be provided to assist in the work of site agent in the site community. At once

At the expected survey meeting, it seems that people get a report of the ARPA program report and works.

There is a considerable interest in the file. In order to solve such a huge storage, the microse film looks very heavy.

want. There are many problems using microsemators, such as only single or limited number of readers, and need hard to copy

Shell equipment, etc. NIC will investigate these problems and start using microsematism materials.

The NIC is using an IMLAC terminal to perform NLS remote access tests. People's graphical access to NLC

A great interest has been generated. The NIC believes that graphics access is not a very high-priority requirement,

However, the programming resources will be provided to the site as soon as possible to test the test of graphical access.

Steve Crocker proposes how people get access to access from different sites and learn to use service facilities

The problem. About users on the network need to include or attach to user documents?

The problem is also discussed. And put forward hard copy, what materials should be included, what are the online information?

question. The NIC will study these issues and generate a list of recommended steps that handle user operations, as well as network access enabling information.

Dick Watson pointed out that although the service provided by NIC can only run a slower terminal and a single layer graphical interface.

Users of NIC may still think that the use of 30 characters per second with a multi-layer graphical interface.

The typing terminal is the best choice. The RFC97 document (NIC5740) describes the most connected to NIC

The initial agreement. Due to the establishment of a committee that produces a standard typewriter connection protocol, the RFC97 file agreement will

Changed to meet a transition agreement recommended by the committee. Recently, with this transition agreement

Submit a new RFC document. Since this meeting, the Type Connection Agreement has decided not to submit this.

Transition protocol.

The meeting was transferred to the file transfer between the site, discussed the transfer of file names through the file in the file transfer without

You need a user to log in to every site. The University of Illinois will submit one on this topic.

An initial RFC document.

Friday morning, February 19

There are some issues discussed at the meeting with network-related data management. The meeting mentioned the following aspects and their

principal:

10 to 12 bits of data machine

Data management language

Form machine

Illiac IV Information Management System

Transition file system

File transfer protocol

The data has been responsible for US Computer Co., Ltd., but requires the Illiac IV information management system and network and in number.

It is based on jobs in management language.

The work of data management languages ​​by the University of Illinois J. Madden, Harvard University Bob Metcalfe,

Rand's J. Heafner, UCSB JIM White and IBM Doug McKay are responsible.

John Heafner said he intends to implement his plan on the form of the company (RFC83 document (NIC5609)

UCSB, MULTICs). Lincoln Lab also claims that they are interested in this. Some sites such as UCLA, SRI,

Rand, Illinois University, Raytheon, Mitre is in UCSB 360/75 within one to three months

Save files on disk arrays are very interested. Jim White said that he will establish one in the next four to six weeks.

A system so that network users can save documents in UCSB.

The problem passed by the host file is again submitted by the University of Illinois. G. Grossman. He said

He will start a dialogue about this topic by generating RFC.

Some people have proposed the problem of username and user ID meaning in the number of interfaces. So far, the number of interfaces is not coming

Configuration, but some people think that some structures such as records, document transfer, and process in processes are valuable. One by one by one

Rand's J. Heafner, MIT-MULTICS E. Meyer and G. Grossman at the University of Illinois

The committee will propose an RFC file to discuss another proposal of the number of interfaces.

UCLA pointed out that it wishes to get a link number in the experimental range of the number of links for the network's test

Test experiment. Link number 223 is assigned to this function. (The link 223 is later discovered and specified, the selected connection

It is actually 191, see RFC104 document (NIC5768)).

Since the machine or system on the network provides service functions, a number of counts is proposed. Current service

It is mainly UCLA's 360/91, UCSB 360/75, SRI's NIC, MIT's Multics, IlliaAc IV,

Lincoln Laboratory 360/67, and data machine. Senior Host ---- Host Agreement Research Committee is beginning to ask

question. They briefly mentioned the network bandwidth problem. BBN's BOB Kahn said he will submit an article to start a discussion about the theme of the count.

Next, a problem with respecting management processes is proposed, such as the number of counts from external systems. Dick

Watson said that he will study and see if the NIC may give help in his solution.

Finally, consider the frequency of the NWG meeting and how to use the problem. Everyone consistently thinks that the NWG meeting is

A very useful meeting, but some preparations for some specific topics to be discussed should be completed in advance. Hope

People who submit them to the subject should be asgeneted as much as possible before the next meeting begins a month. PEGGY

KARP will be responsible for NWG's room problem in the spring meeting. She will soon issue a reservation request.

[This RFC document is marked by Kelly Tardif, Viagie in October 1999]

[Compose the machine readable form to entered the RFC online file]

RFC0101 --- Notes on The Network Working Group Meeting GROUP

Comment on the Note on the Network Working Group of Urbana in Illinois on February 17, 1971

8

RFC Document Chinese Translation Program

转载请注明原文地址:https://www.9cbs.com/read-127628.html

New Post(0)