----MESSAGE-BEGIN---- [anews.Aucbvax.4889] <1981110117162700> Message-ID: Newsgroups: fa.tcp-ip X-Path: utzoo!decvax!ucbvax!tcp-ip From: ucbvax!tcp-ip Date: Sun Nov 1 22:16:27 1981 Subject: TCP-IP Digest, Vol 1 #5 X-Google-Info: Converted from the original A-News header >From tcp-ip@brl Sun Nov 1 21:56:45 1981 TCP/IP Digest Sunday, 1 November 1981 Volume 1 : Issue 5 Today's Topics: Solicitation for More Contributions More on BBN TCP for TOPS-20 Z8000 based TCP and IP Front End Machine ---------------------------------------------------------------------- From: Mike Muuss Reply-to: tcp-ip@brl Subject: More Contributions? Nearly a week has passed since the last issue, so I am publishing the three letters that have arrived in the interim . Considering the size of the mailing list, I can hardly immagine that we have heard from everybody who is working on networking implementations. C'mon! Lets hear from everybody. Cheers, -Mike ------------------------------ Date: 27 Oct 1981 1810-EST From: Jonathan Alan Solomon Subject: List maintainence You can announce [Rutgers]MAIL.TXT as an archive point if you like. ------------------------------ Date: 26 Oct 1981 0013-PST From: Mark Crispin Postal-Address: 725 Mariposa Ave. #103; Mountain View, CA 94041 Phone: (415) 497-1407 (Stanford); (415) 968-1052 (residence) Subject: Re: TCP-IP Digest, Vol 1 #4 I'd like to answer a few confusions about my position regarding the TOPS-20 implementation of TCP available from BBN. I am not, nor have I ever been, opposed to the TCP protocol. I was very impressed with the work done at the DoD Protocol Standards conference a year ago. I've been urging the managers of the Stanford local network effort to adopt TCP/IP as the local network protocol for the past two years now. It is the software that is presently available for TOPS-20 that I find distasteful. I have had some preliminary discussions with various people at DEC about this issue, and I have determined that they are addressing at least some of the objections. If the product DEC releases is less than what we would like, it is because of their rush to meet the deadline. It's a safe assumption that there is no way that DEC can possibly have a rewritten TCP implementation for TOPS-20 out in the field by the deadline date. I believe that DEC is doing its best. DEC's customers are probably best off encouraging the current project but being firm in stating that we must have a rewrite which addresses the performance problems of BBN's TCP. So far as the comments on how to "help/force people [to] implement TCP/IP" go: (1) There are those of us who would feel that not being able to reach our systems from a TIP is a feature, and not a problem at all! Entirely too many high school kids abuse the network from TIPs. (2) "Getting the mail through" can be accomplished by other means than implementing TCP. (3) Services only accessible via TCP/IP are a good reason for implementing TCP/IP. The example given was not a good one, but I can see other valuable resources being TCP/IP only. I hope by the time such resources exist there will be a better implementation of TCP/IP available. (4) The last point is patently ridiculous. US mail existed before electronic mail, and is still a commonly used method for communication between Investigators and their Sponsors. The whole tone of "forcing" is itself inane. The intent of my message was to discuss getting things moving towards solving the software situation, not to create an anti-TCP/IP lobby. The present TCP/IP software for TOPS-20 is unpalatable for most sites; if "forced" to implement TCP/IP on our systems we will probably have to write the software ourselves. Of course that would keep us from completing the projects our Network Sponsors are supporting us to do... -- Mark -- ------------------------------ Date: Mon, 26 Oct 1981 2236-EST From: steve at MITRE Subject: Z8000 TCP and IP Mike, I have been watching your digest with great interest. We have produced a micro-based TCP/IP here at MITRE which your readers may find useful. We have been involved in a series of projects whose focus was to make network access (both local and long haul) as straightforward as using common computer peripherals. The idea was that just as hardware controllers handle the particulars of a disk cylinder centerline or an end of tape sensor, some of the new microprocessors, in our case the Zilog Z8000, should make it possible to construct a "network controller" to handle the par- ticulars of packet ordering and flow control. To a large extent we have succeeded with a TCP/IP network controller supported by a Z8000 microprocessor box. The box is currently interfaced to our UNIX system via a UMC-Z80. It is accessed from the users point of view as a set of I/O like management calls (open, close, read, write, and special) which are transported via a network access protocol to the outboard box. The box has 64K bytes of Ram, 32K bytes of Rom, a Z8002 micro, and a serial Usart (880K BPS max). All of the software was written in C using a locally brewed version of the portable C compiler. The interesting aspect of the box is that it inter- faces as easily to a local network (in our case a the MITRE RF cable backbone) as it does to the ARPA network. Other than dif- ferent round trip delays, host user-level software sees no dis- tinction between the two networks. The long haul metamorphosis involves a new device driver in the Z8000 and the addition of an ACC 1822 hardware interface (yes, ACC makes one). The resulting set of building blocks allows us to interface a host, a terminal concentrator version and other units to a local net and have a gateway to the arpanet. While a custom protocol would be faster, we believe that the longer term interoperability of TCP/IP will be well worth some short term overhead. The performance even with TCP/IP isn't that bad in that we have measured two user processes talking via TCP/IP over the cable at 350K BPS. We have measured rates as high at 450K BPS when user I/O buffer sizes are set at 8K bytes per I/O. The Internet Protocol contains our lowest level of addressing. This allows us to address local units in the same way we address remote units two or three networks away. We have been experimenting with a version of TCP/IP which allows the optional specification of some TCP and IP mechanisms. The basic conclusion is that cable signalling rates are so fast that the effect of 300 bit TCP/IP headers has negligible impact on perfor- mance. Our work this year involves constructing a new version 10M BPS controller with multi-microprocessor capabilities. We believe the resulting effective TCP/IP communications rates should be well above 1M BPS and that the multi-microprocessor capabilities should make for an interesting distributed process- ing base. There a couple of reports available if people are in- terested. Regards Steve Holmgren END OF TCP-IP DIGEST ******************** ----MESSAGE-END---- ----MESSAGE-BEGIN---- <1981110119185600> Mail-from: ARPANET host BRL rcvd at 1-Nov-81 2238-PST Date: 1 Nov 81 23:18:56-EDT (Sun) From: Mike Muuss To: tcp-ip at Brl Subject: TCP-IP Digest, Vol 1 #5 TCP/IP Digest Sunday, 1 November 1981 Volume 1 : Issue 5 Today's Topics: Solicitation for More Contributions More on BBN TCP for TOPS-20 Z8000 based TCP and IP Front End Machine ---------------------------------------------------------------------- From: Mike Muuss Reply-to: tcp-ip@brl Subject: More Contributions? Nearly a week has passed since the last issue, so I am publishing the three letters that have arrived in the interim . Considering the size of the mailing list, I can hardly immagine that we have heard from everybody who is working on networking implementations. C'mon! Lets hear from everybody. Cheers, -Mike ------------------------------ Date: 27 Oct 1981 1810-EST From: Jonathan Alan Solomon Subject: List maintainence You can announce [Rutgers]MAIL.TXT as an archive point if you like. ------------------------------ Date: 26 Oct 1981 0013-PST From: Mark Crispin Postal-Address: 725 Mariposa Ave. #103; Mountain View, CA 94041 Phone: (415) 497-1407 (Stanford); (415) 968-1052 (residence) Subject: Re: TCP-IP Digest, Vol 1 #4 I'd like to answer a few confusions about my position regarding the TOPS-20 implementation of TCP available from BBN. I am not, nor have I ever been, opposed to the TCP protocol. I was very impressed with the work done at the DoD Protocol Standards conference a year ago. I've been urging the managers of the Stanford local network effort to adopt TCP/IP as the local network protocol for the past two years now. It is the software that is presently available for TOPS-20 that I find distasteful. I have had some preliminary discussions with various people at DEC about this issue, and I have determined that they are addressing at least some of the objections. If the product DEC releases is less than what we would like, it is because of their rush to meet the deadline. It's a safe assumption that there is no way that DEC can possibly have a rewritten TCP implementation for TOPS-20 out in the field by the deadline date. I believe that DEC is doing its best. DEC's customers are probably best off encouraging the current project but being firm in stating that we must have a rewrite which addresses the performance problems of BBN's TCP. So far as the comments on how to "help/force people [to] implement TCP/IP" go: (1) There are those of us who would feel that not being able to reach our systems from a TIP is a feature, and not a problem at all! Entirely too many high school kids abuse the network from TIPs. (2) "Getting the mail through" can be accomplished by other means than implementing TCP. (3) Services only accessible via TCP/IP are a good reason for implementing TCP/IP. The example given was not a good one, but I can see other valuable resources being TCP/IP only. I hope by the time such resources exist there will be a better implementation of TCP/IP available. (4) The last point is patently ridiculous. US mail existed before electronic mail, and is still a commonly used method for communication between Investigators and their Sponsors. The whole tone of "forcing" is itself inane. The intent of my message was to discuss getting things moving towards solving the software situation, not to create an anti-TCP/IP lobby. The present TCP/IP software for TOPS-20 is unpalatable for most sites; if "forced" to implement TCP/IP on our systems we will probably have to write the software ourselves. Of course that would keep us from completing the projects our Network Sponsors are supporting us to do... -- Mark -- ------------------------------ Date: Mon, 26 Oct 1981 2236-EST From: steve at MITRE Subject: Z8000 TCP and IP Mike, I have been watching your digest with great interest. We have produced a micro-based TCP/IP here at MITRE which your readers may find useful. We have been involved in a series of projects whose focus was to make network access (both local and long haul) as straightforward as using common computer peripherals. The idea was that just as hardware controllers handle the particulars of a disk cylinder centerline or an end of tape sensor, some of the new microprocessors, in our case the Zilog Z8000, should make it possible to construct a "network controller" to handle the par- ticulars of packet ordering and flow control. To a large extent we have succeeded with a TCP/IP network controller supported by a Z8000 microprocessor box. The box is currently interfaced to our UNIX system via a UMC-Z80. It is accessed from the users point of view as a set of I/O like management calls (open, close, read, write, and special) which are transported via a network access protocol to the outboard box. The box has 64K bytes of Ram, 32K bytes of Rom, a Z8002 micro, and a serial Usart (880K BPS max). All of the software was written in C using a locally brewed version of the portable C compiler. The interesting aspect of the box is that it inter- faces as easily to a local network (in our case a the MITRE RF cable backbone) as it does to the ARPA network. Other than dif- ferent round trip delays, host user-level software sees no dis- tinction between the two networks. The long haul metamorphosis involves a new device driver in the Z8000 and the addition of an ACC 1822 hardware interface (yes, ACC makes one). The resulting set of building blocks allows us to interface a host, a terminal concentrator version and other units to a local net and have a gateway to the arpanet. While a custom protocol would be faster, we believe that the longer term interoperability of TCP/IP will be well worth some short term overhead. The performance even with TCP/IP isn't that bad in that we have measured two user processes talking via TCP/IP over the cable at 350K BPS. We have measured rates as high at 450K BPS when user I/O buffer sizes are set at 8K bytes per I/O. The Internet Protocol contains our lowest level of addressing. This allows us to address local units in the same way we address remote units two or three networks away. We have been experimenting with a version of TCP/IP which allows the optional specification of some TCP and IP mechanisms. The basic conclusion is that cable signalling rates are so fast that the effect of 300 bit TCP/IP headers has negligible impact on perfor- mance. Our work this year involves constructing a new version 10M BPS controller with multi-microprocessor capabilities. We believe the resulting effective TCP/IP communications rates should be well above 1M BPS and that the multi-microprocessor capabilities should make for an interesting distributed process- ing base. There a couple of reports available if people are in- terested. Regards Steve Holmgren END OF TCP-IP DIGEST ******************** ----MESSAGE-END---- ----MESSAGE-BEGIN---- [anews.Aucbvax.5236] <1981111800501900> Message-ID: Newsgroups: fa.tcp-ip X-Path: utzoo!decvax!ucbvax!tcp-ip From: ucbvax!tcp-ip Date: Wed Nov 18 05:50:19 1981 Subject: TCP-IP Digest, Vol 1 #6 X-Google-Info: Converted from the original A-News header >From tcp-ip@brl Wed Nov 18 05:33:31 1981 TCP/IP Digest Wednesday, 11 Nov 1981 Volume 1 : Issue 6 Today's Topics: Administrivia: Long Delay Disabling NCPs -- Directory Service? TOPS-20 TCP Defenses? -- TCP/IP for Cybers? Unix/ArpaNet TCP Problems & Solutions TCP/IP Performance on VAXen Berkeley Enhanced TCP/IP for 4BSD UNIX ---------------------------------------------------------------------- From: Mike Muuss Subject: Administrivia Hello again folks! Sorry about the long delay between this digest and the previous one -- I was in Denver on travel, and could not get to a terminal. I hope to send out future digests as soon as enough material has been accumulated. Cheers, -Mike ------------------------------ From: POSTEL at USC-ISIF Subject: Disabling NCPs There has been some talk of "forcing" the move to TCP by various administrative and policy measures. There was also a claim that there was no technical way to force the abandonment of NCP. It should be pointed out that a quite simple modification to the IMP program would enable the IMPs to filter out and discard all NCP traffic. As far as i know, there has been no decision to do this, but you should be aware that it is technical feasible. --jon. ------------------------------ From: Harris A. Meyers Subject: Request A useful fuction for this list to perform might be to produce a directory of who has IP/TCP available for what machines and operating systems. In particular I am looking for a version to run on IBM's VM/CMS. harris ------------------------------ Subject: Re: TOPS 20 TCP/IP From: mike at RAND-UNIX I have often heard criticisms of TOPS 20 TCP/IP implementation, but never a defense. Does anyone from BBN or ARPA care to defend their implementation or do they agree with the criticisms? ------------------------------ From: roya at UTEXAS-11 Subject: Is there TCP/IP for Cyber machines? Do any of you know any sites that might have or is planning to implement TCP/IP for CDC Cyber machines like 175 with NOS like operating system? Roya [ Tektronix Labs (Clem Cole et.al.) have already implemented TCP and IP in RatFor for their Cyber, running under NOS. -Mike ] ------------------------------ From: greg at NPRDC Subject: Future contributions Now that all the special interest groups have spoken, can we get back to the original subject? In case you've forgotten, it was "Unix/ARPAnet TCP problems and solutions." Although I'm interested in the various problems/ possibilities of using TCP on other operating systems or other ethers, at a minimum, our mutual interest is getting our machines running TCP before the deadline. (Probabally this list goes a little farther than that; to those people, I appologize. But we are the ones with the deadline fast approaching.) Maybe we can discuss theoretical issues later, but I am more interested in the practical issues -- namely, who has TCP up? How is it connected to the ARPAnet (or even another ether, if the problems/ solutions are similar)? What problems were encountered? How fast is it? How does it compare in simplicity/performance/transparancy/completeness/ functionality/limitations/etc. with the other possibilities? So far, we have heard of two real choices (assuming that we're not going to have to buy any additional hardware): BBN and 3COM. Who's got them up? How connected? (I am VDH, so solutions that don't have a VDH driver are uninteresting.) Speak up; now's your chance to brag, and you can do the rest of us a real service. [ Actually, I had hoped that this digest could serve as a forum for technical discussion of networking for ALL systems, but clearly the transition to TCP for current ArpaNet Hosts is the primary motivator. I hope that this list will not restrict itself just to UNIX, though. -Mike] ------------------------------ Subject: TCP/IP Performance on VAX From: CERF at USC-ISI UC Berkeley has been developing a paging UNIX(TM) for the VAX based on V7 UNIX (TM-Western Electric). BBN has been developing a TCP/IP for this VAX UNIX(TM) and UCB recently reported data bandwidths of Mb/sec over a 3 Mb/s Ethernet running TCP/IP to TCP/IP including checksumming. This figure obtained on VAX 11/750 using 1 kilobyte packets. A point of contact for the Berkeley + BBN VAX UNIX(TM) is Prof. Robert Fabry (ARPAVAX.Fabry@Berkeley). A point of contact at BBN is Gurwitz@BBN-RSM(Rob Gurwitz). Vint Cerf (bd) ------------------------------ From: ARPAVAX .wnj at Berkeley Subject: tcp-ip digest contribution Cc: gurwitz@bbn-unix Moderator: Here is a description of the work we are doing a Berkeley with tcp-ip. I hope it is in time for this weeks digest. We have enjoyed the past digests and hope that future digests will be as interesting. Regards, Bill Joy ===== begin ===== The Computer Systems Research Group at Berkeley is enhancing the UNIX operating system with DARPA support. We are improving UNIX memory management facilities, working on extensions to UNIX to support better inter-process communication, and incorporating support for both local and long haul networks. In particular, we expect to try using the INTERNET protocols on a number of different commercially available local network interfaces. The basis for our INTERNET protocols is the TCP/IP implementation done by Rob Gurwitz at BBN. While this TCP has more than adequate performance for use in the ARPANET context, we need extremely good performance to be able to use the protocols as the basis for construction of distributed UNIX applications between machines on a local network. In particular, we wish to insure that data can be transferred rapidly between VAX machines on 10 Megabaud Ethernet cables. Our current file system organization uses 1024 byte records, while a newer, high-performance file system will use 4096 byte records. We are therefore interested in the effective transmission of records of these sizes. We are also interested in low per-packet overhead for distributed applications which exchange many messages. We have just finished about three weeks of tuning of the BBN TCP/IP for our 3 Megabaud prototype Ethernet. We had previously brought TCP/IP up on the Ethernet and were interested in learning more about the internals of TCP and discovering whether the protocol would be a bottleneck when running on a local network. The results we have obtained suggest that this is not the case. As an experiment to investigate the performance of the resulting TCP/IP implementation, we transmitted 4 Megabytes of data between two user processes on different machines. The transfer was partitioned into 1024 byte records and encapsulated in 1068 byte Ethernet packets. Sending the data from our 11/750 to our 11/780 through TCP/IP takes 28 seconds. This includes all the time needed to set up and tear down the connections, for an user-user throughput of 1.2 Megabaud. During this time the 11/750 is CPU saturated, but the 11/780 has about 30% idle time. The time spent in the system processing the data is spread out among handling for the Ethernet (20%), IP packet processing (10%), TCP processing (30%), checksumming (25%), and user system call handling (15%), with no single part of the handling dominating the time in the system. The TCP performance exceeds the throughput of the current VAX/UNIX file system by a factor of 3, due to the small block size of the UNIX file system. It comes within a factor of 2 of our per-spindle performance of a early prototype of a new file system organization we are working on. The relative speed of the TCP/IP protocol and the file system suggests that we will be able to transfer volumes of data regularly between machines without any special protocols. The limited bandwidth of our 3 Megabaud cable may be a bottleneck until we put in one or more 10 Megabaud cables. Higher rates can be expected between VAX-11/780's, but we have no direct measurements yet, as we have only one 780 with easily accessible down time. Improvements are yet possible within the bounds of the TCP/IP protocol and the new Ethernet standard (which limits packets to about 1500 bytes). In particular, we may use IP fragmentation and reassembly on the local network to allow the TCP and higher level system code to process 4096 byte data records (which is a more natural block size for the newer system we are working on, being a basic file system data page size.) This is convenient within the bounds of the Ethernet standards only because IP supports fragmentation and reassembly in a general way. Simple techniques are also available which would reduce the number of ACK's by a significant amount to further speed the TCP. Using information gathered from UNIX kernel profiling we can estimate the speed improvements possible given a 10 Megabaud cable. (All of these projections will be for user-user throughput between 11/750's.) Switching to 4096 byte segments in TCP transfers, while fragmenting at the IP layer (to stay within the packet length restriction of the Ethernet standard) we estimate an increased throughput of 1.9 Megabaud. This is with an interface which is functionality equivalent to our current prototype Ethernets. We plan on changing UNIX soon so that user i/o buffers are naturally page aligned. This would eliminate a copy of the data (when it is read) to raise the throughput to about 2.25 Megabaud. (This speedup only at the receiver gives an end-end speedup because the receiving process otherwise has more work to do.) With checksum calculation support from the interface hardware the user-user bandwidth would rise to about 3.5 Megabaud. At this point the major overhead is the processing of the 4 interrupts for the fragments of the 4k packets; a transmission medium which allowed us to use 4k packets would let the bandwidth rise to about 6.6 Megabaud, user-user. A final improvement would be to implement a variant on the send system call which released the virtual memory when the message was sent. This would be very useful for servers and could also be used by the standard i/o library. This would reduce the transmit overhead (which at this point should be greater than the receive overhead) making the final throughput about 8 Megabaud. On 11/780's, these numbers typically scale up by 1.4 so that we can project the throughput with the improvements described above to be about 11.2 Megabaud, user-user. While factors in the VAX architecture other than the protocol might well dominate before this bandwidth is achieved, this means that high data rates through the protocol can be achieved with a relatively small percentage of the processor. We are working on IPC facilities for UNIX which will interface to the INTERNET protocol family, and allow us to construct distributed applications for UNIX without explicit dependence on the network layer protocols. The measurements reported here suggest that we do not need to look for more efficient local network protocols, but will need to support other protocols only for inter-operability with other networks and systems. We will be working with Rob Gurwitz at BBN in the coming weeks, combining our version of TCP/IP with his current version. We look forward to making a high-performance version of the protocol available to the VAX/UNIX community at an early date. Regards, Bill Joy and Bob Fabry End of TCP-IP Digest ******************** ----MESSAGE-END---- ----MESSAGE-BEGIN---- <1981111801395900> Mail-from: ARPANET host USC-ISIF rcvd at 18-Nov-81 0915-PST Mail-from: ARPANET host BRL rcvd at 18-Nov-81 0905-PST Date: 18 Nov 81 5:39:59-EDT (Wed) From: Mike Muuss To: list: Subject: TCP-IP Digest, Vol 1 #6 TCP/IP Digest Wednesday, 11 Nov 1981 Volume 1 : Issue 6 Today's Topics: Administrivia: Long Delay Disabling NCPs -- Directory Service? TOPS-20 TCP Defenses? -- TCP/IP for Cybers? Unix/ArpaNet TCP Problems & Solutions TCP/IP Performance on VAXen Berkeley Enhanced TCP/IP for 4BSD UNIX ---------------------------------------------------------------------- From: Mike Muuss Subject: Administrivia Hello again folks! Sorry about the long delay between this digest and the previous one -- I was in Denver on travel, and could not get to a terminal. I hope to send out future digests as soon as enough material has been accumulated. Cheers, -Mike ------------------------------ From: POSTEL at USC-ISIF Subject: Disabling NCPs There has been some talk of "forcing" the move to TCP by various administrative and policy measures. There was also a claim that there was no technical way to force the abandonment of NCP. It should be pointed out that a quite simple modification to the IMP program would enable the IMPs to filter out and discard all NCP traffic. As far as i know, there has been no decision to do this, but you should be aware that it is technical feasible. --jon. ------------------------------ From: Harris A. Meyers Subject: Request A useful fuction for this list to perform might be to produce a directory of who has IP/TCP available for what machines and operating systems. In particular I am looking for a version to run on IBM's VM/CMS. harris ------------------------------ Subject: Re: TOPS 20 TCP/IP From: mike at RAND-UNIX I have often heard criticisms of TOPS 20 TCP/IP implementation, but never a defense. Does anyone from BBN or ARPA care to defend their implementation or do they agree with the criticisms? ------------------------------ From: roya at UTEXAS-11 Subject: Is there TCP/IP for Cyber machines? Do any of you know any sites that might have or is planning to implement TCP/IP for CDC Cyber machines like 175 with NOS like operating system? Roya [ Tektronix Labs (Clem Cole et.al.) have already implemented TCP and IP in RatFor for their Cyber, running under NOS. -Mike ] ------------------------------ From: greg at NPRDC Subject: Future contributions Now that all the special interest groups have spoken, can we get back to the original subject? In case you've forgotten, it was "Unix/ARPAnet TCP problems and solutions." Although I'm interested in the various problems/ possibilities of using TCP on other operating systems or other ethers, at a minimum, our mutual interest is getting our machines running TCP before the deadline. (Probabally this list goes a little farther than that; to those people, I appologize. But we are the ones with the deadline fast approaching.) Maybe we can discuss theoretical issues later, but I am more interested in the practical issues -- namely, who has TCP up? How is it connected to the ARPAnet (or even another ether, if the problems/ solutions are similar)? What problems were encountered? How fast is it? How does it compare in simplicity/performance/transparancy/completeness/ functionality/limitations/etc. with the other possibilities? So far, we have heard of two real choices (assuming that we're not going to have to buy any additional hardware): BBN and 3COM. Who's got them up? How connected? (I am VDH, so solutions that don't have a VDH driver are uninteresting.) Speak up; now's your chance to brag, and you can do the rest of us a real service. [ Actually, I had hoped that this digest could serve as a forum for technical discussion of networking for ALL systems, but clearly the transition to TCP for current ArpaNet Hosts is the primary motivator. I hope that this list will not restrict itself just to UNIX, though. -Mike] ------------------------------ Subject: TCP/IP Performance on VAX From: CERF at USC-ISI UC Berkeley has been developing a paging UNIX(TM) for the VAX based on V7 UNIX (TM-Western Electric). BBN has been developing a TCP/IP for this VAX UNIX(TM) and UCB recently reported data bandwidths of Mb/sec over a 3 Mb/s Ethernet running TCP/IP to TCP/IP including checksumming. This figure obtained on VAX 11/750 using 1 kilobyte packets. A point of contact for the Berkeley + BBN VAX UNIX(TM) is Prof. Robert Fabry (ARPAVAX.Fabry@Berkeley). A point of contact at BBN is Gurwitz@BBN-RSM(Rob Gurwitz). Vint Cerf (bd) ------------------------------ From: ARPAVAX .wnj at Berkeley Subject: tcp-ip digest contribution Cc: gurwitz@bbn-unix Moderator: Here is a description of the work we are doing a Berkeley with tcp-ip. I hope it is in time for this weeks digest. We have enjoyed the past digests and hope that future digests will be as interesting. Regards, Bill Joy ===== begin ===== The Computer Systems Research Group at Berkeley is enhancing the UNIX operating system with DARPA support. We are improving UNIX memory management facilities, working on extensions to UNIX to support better inter-process communication, and incorporating support for both local and long haul networks. In particular, we expect to try using the INTERNET protocols on a number of different commercially available local network interfaces. The basis for our INTERNET protocols is the TCP/IP implementation done by Rob Gurwitz at BBN. While this TCP has more than adequate performance for use in the ARPANET context, we need extremely good performance to be able to use the protocols as the basis for construction of distributed UNIX applications between machines on a local network. In particular, we wish to insure that data can be transferred rapidly between VAX machines on 10 Megabaud Ethernet cables. Our current file system organization uses 1024 byte records, while a newer, high-performance file system will use 4096 byte records. We are therefore interested in the effective transmission of records of these sizes. We are also interested in low per-packet overhead for distributed applications which exchange many messages. We have just finished about three weeks of tuning of the BBN TCP/IP for our 3 Megabaud prototype Ethernet. We had previously brought TCP/IP up on the Ethernet and were interested in learning more about the internals of TCP and discovering whether the protocol would be a bottleneck when running on a local network. The results we have obtained suggest that this is not the case. As an experiment to investigate the performance of the resulting TCP/IP implementation, we transmitted 4 Megabytes of data between two user processes on different machines. The transfer was partitioned into 1024 byte records and encapsulated in 1068 byte Ethernet packets. Sending the data from our 11/750 to our 11/780 through TCP/IP takes 28 seconds. This includes all the time needed to set up and tear down the connections, for an user-user throughput of 1.2 Megabaud. During this time the 11/750 is CPU saturated, but the 11/780 has about 30% idle time. The time spent in the system processing the data is spread out among handling for the Ethernet (20%), IP packet processing (10%), TCP processing (30%), checksumming (25%), and user system call handling (15%), with no single part of the handling dominating the time in the system. The TCP performance exceeds the throughput of the current VAX/UNIX file system by a factor of 3, due to the small block size of the UNIX file system. It comes within a factor of 2 of our per-spindle performance of a early prototype of a new file system organization we are working on. The relative speed of the TCP/IP protocol and the file system suggests that we will be able to transfer volumes of data regularly between machines without any special protocols. The limited bandwidth of our 3 Megabaud cable may be a bottleneck until we put in one or more 10 Megabaud cables. Higher rates can be expected between VAX-11/780's, but we have no direct measurements yet, as we have only one 780 with easily accessible down time. Improvements are yet possible within the bounds of the TCP/IP protocol and the new Ethernet standard (which limits packets to about 1500 bytes). In particular, we may use IP fragmentation and reassembly on the local network to allow the TCP and higher level system code to process 4096 byte data records (which is a more natural block size for the newer system we are working on, being a basic file system data page size.) This is convenient within the bounds of the Ethernet standards only because IP supports fragmentation and reassembly in a general way. Simple techniques are also available which would reduce the number of ACK's by a significant amount to further speed the TCP. Using information gathered from UNIX kernel profiling we can estimate the speed improvements possible given a 10 Megabaud cable. (All of these projections will be for user-user throughput between 11/750's.) Switching to 4096 byte segments in TCP transfers, while fragmenting at the IP layer (to stay within the packet length restriction of the Ethernet standard) we estimate an increased throughput of 1.9 Megabaud. This is with an interface which is functionality equivalent to our current prototype Ethernets. We plan on changing UNIX soon so that user i/o buffers are naturally page aligned. This would eliminate a copy of the data (when it is read) to raise the throughput to about 2.25 Megabaud. (This speedup only at the receiver gives an end-end speedup because the receiving process otherwise has more work to do.) With checksum calculation support from the interface hardware the user-user bandwidth would rise to about 3.5 Megabaud. At this point the major overhead is the processing of the 4 interrupts for the fragments of the 4k packets; a transmission medium which allowed us to use 4k packets would let the bandwidth rise to about 6.6 Megabaud, user-user. A final improvement would be to implement a variant on the send system call which released the virtual memory when the message was sent. This would be very useful for servers and could also be used by the standard i/o library. This would reduce the transmit overhead (which at this point should be greater than the receive overhead) making the final throughput about 8 Megabaud. On 11/780's, these numbers typically scale up by 1.4 so that we can project the throughput with the improvements described above to be about 11.2 Megabaud, user-user. While factors in the VAX architecture other than the protocol might well dominate before this bandwidth is achieved, this means that high data rates through the protocol can be achieved with a relatively small percentage of the processor. We are working on IPC facilities for UNIX which will interface to the INTERNET protocol family, and allow us to construct distributed applications for UNIX without explicit dependence on the network layer protocols. The measurements reported here suggest that we do not need to look for more efficient local network protocols, but will need to support other protocols only for inter-operability with other networks and systems. We will be working with Rob Gurwitz at BBN in the coming weeks, combining our version of TCP/IP with his current version. We look forward to making a high-performance version of the protocol available to the VAX/UNIX community at an early date. Regards, Bill Joy and Bob Fabry End of TCP-IP Digest ******************** ----MESSAGE-END---- ----MESSAGE-BEGIN---- [anews.Aucbvax.5333] <1981112906195600> Message-ID: Newsgroups: fa.tcp-ip X-Path: utzoo!decvax!ucbvax!tcp-ip From: ucbvax!tcp-ip Date: Sun Nov 29 11:19:56 1981 Subject: TCP-IP Digest, Vol 1 #7 X-Google-Info: Converted from the original A-News header >From tcp-ip@brl Sun Nov 29 10:12:00 1981 TCP/IP Digest Sunday, 29 Nov 1981 Volume 1 : Issue 7 Today's Topics: Query: TCP-IP for RSX-11M? List of TCP-Capable Hosts UUCP Networking Concerns + Internet Issues General Networking Issues "Fair Game" in TCP-IP Digest Query: TCP-IP for NOS/BE? ---------------------------------------------------------------------- From: RCOLE at USC-ISIE Subject: for TCP-IP digest I would like to obtain information from anyone who has (or knows of) a TCP-IP implementation for RSX11-M, the machine in question is an 11/70. Any help gratefully received. Reply to: Robert Cole, University College, London. RCOLE@ISIE ------------------------------ Subject: Re: TCP From: CERF at USC-ISI To: MCCUNE at USC-ISIC Cc: tcp-ip at BRL [TCP] TELNET is available on all ISI hosts; FTP is available on the VAX and is being worked on for the TOPS-20 by BBN (Charles Lynn). A list of TCP-capable hosts is in [ISIF]TCP-IP-STATUS.TXT. TN runs on the TOPS-20 ("TelNet") and can be used to go from ISIC to the VAX. Vint The file at ISIF is in need of updating - Jon Postel will be pulsing the TCP implementers for new info, but you may find the current version of some interst. ------------------------------ From: chico!duke!unc!smb at Berkeley In-real-life: Steven M. Bellovin Subject: Re: systems news articles From duke!dbl Wed Nov 18 14:01:41 1981 Date-Sent: Wed Nov 18 13:57:35 1981 To: unc!smb Subject: systems news articles MMDF- swd has done some looking into it as well, and thinks it's the way to go. We've heard from farber and crocker at udel that we will probably be the next csnet site to go online, so we'll be getting the most up-to-date version of MMDF sometime this month (or maybe early December). I'm rapidly coming to that conclusion as well, and the sooner I see some manuals and some code, the happier I'll be. I just read RFC754 and RFC799, and it's becoming apparent that the ARPAnuts are setting standards which we'll have to adhere to if we're to talk to them. And the whole uucp addressing mess is getting out of hand -- and that says nothing of changing topologies (what do we do with wolfvax's mail when they hook in to MCNC instead) or the distinction users here have to make between pnet mail and uucp mail. Add in ARPA, CSnet, and maybe Berknet among the duke machines, and you have a royal mess. I'm inclined to start a new net newsgroup to discuss mail, networking, addressing, etc., from a UNIX/uucp point of view -- say, net.net (fa.tcp-ip appears to be too specialized, though I'll route a copy of this to the moderator). ------------------------------ Date: 19 Nov 1981 07:40:13-PST From: cbosgd!mark at Berkeley Subject: Re: systems news articles Having a newsgroup to discuss nets is different than discussing mail. I propose net.net and net.mail. I'm not sure net.net is needed - does fa.tcp-ip subsume it? There will probably soon be a net.csnet, too. Mark ------------------------------ From: Mike Muuss To: chico!duke!unc!smb at Ucb-C70 Subject: Networking Discussions in TCP-IP Digest? Steve - While the masthead "TCP-IP Digest" is really rather specialized, I had intended the Digest as more of a discussion on IMPLEMENTATION issues of networking (as opposed to Philosophical discussions as get found in HUMAN-NETS). The troubles with multiple networks, and the variety of message formats (for mail), and routing problems in general are all fair game for the TCP-IP digest. You are welcome to have this networking discussion in the TCP digest -- if the volume becomes too great I would be willing to clone a new digest later on. BRL polls Duke via UUCP, so messages addressed to ...!duke!bmd70!tcp-ip should make it to the digest (no need to go through Berkeley). Give it a try. Our RMAIL is smart enough to prevent accidental gatewaying; sorry. Cheers, -Mike ------------------------------ Sender: DRDAR-MST at OFFICE-8 Subject: TCP/IP and the CYBERS From: Richard Sitnik In a recent message Mike Muuss mentioned that there are plans to install a version of TCP-IP on the CYBER. Also, in Issue 6 of the TCP-IP Digest Mike mentioned that Tektronix has implemented TCP under NOS. I would like to find out more about plans regarding TCP-IP. In particular, do you know of anyone who is working on a NOS/BE version? I have spoken to Tektronix and their implementation will only work with NOS and requires a HYPER channel. They are not aware of any NOS/BE efforts. I will prepare an item for the TCP-IP Digest detailing my discussion with Tektronix. RICH END OF TCP-IP DIGEST ******************** ----MESSAGE-END---- ----MESSAGE-BEGIN---- <1981112907272900> Mail-from: ARPANET host BRL rcvd at 29-Nov-81 1048-PST Date: 29 Nov 81 11:27:29-EDT (Sun) From: Mike Muuss To: list: Subject: TCP-IP Digest, Vol 1 #7 Bcc: TCP/IP Digest Sunday, 29 Nov 1981 Volume 1 : Issue 7 Today's Topics: Query: TCP-IP for RSX-11M? List of TCP-Capable Hosts UUCP Networking Concerns + Internet Issues General Networking Issues "Fair Game" in TCP-IP Digest Query: TCP-IP for NOS/BE? ---------------------------------------------------------------------- From: RCOLE at USC-ISIE Subject: for TCP-IP digest I would like to obtain information from anyone who has (or knows of) a TCP-IP implementation for RSX11-M, the machine in question is an 11/70. Any help gratefully received. Reply to: Robert Cole, University College, London. RCOLE@ISIE ------------------------------ Subject: Re: TCP From: CERF at USC-ISI To: MCCUNE at USC-ISIC Cc: tcp-ip at BRL [TCP] TELNET is available on all ISI hosts; FTP is available on the VAX and is being worked on for the TOPS-20 by BBN (Charles Lynn). A list of TCP-capable hosts is in [ISIF]TCP-IP-STATUS.TXT. TN runs on the TOPS-20 ("TelNet") and can be used to go from ISIC to the VAX. Vint The file at ISIF is in need of updating - Jon Postel will be pulsing the TCP implementers for new info, but you may find the current version of some interst. ------------------------------ From: chico!duke!unc!smb at Berkeley In-real-life: Steven M. Bellovin Subject: Re: systems news articles From duke!dbl Wed Nov 18 14:01:41 1981 Date-Sent: Wed Nov 18 13:57:35 1981 To: unc!smb Subject: systems news articles MMDF- swd has done some looking into it as well, and thinks it's the way to go. We've heard from farber and crocker at udel that we will probably be the next csnet site to go online, so we'll be getting the most up-to-date version of MMDF sometime this month (or maybe early December). I'm rapidly coming to that conclusion as well, and the sooner I see some manuals and some code, the happier I'll be. I just read RFC754 and RFC799, and it's becoming apparent that the ARPAnuts are setting standards which we'll have to adhere to if we're to talk to them. And the whole uucp addressing mess is getting out of hand -- and that says nothing of changing topologies (what do we do with wolfvax's mail when they hook in to MCNC instead) or the distinction users here have to make between pnet mail and uucp mail. Add in ARPA, CSnet, and maybe Berknet among the duke machines, and you have a royal mess. I'm inclined to start a new net newsgroup to discuss mail, networking, addressing, etc., from a UNIX/uucp point of view -- say, net.net (fa.tcp-ip appears to be too specialized, though I'll route a copy of this to the moderator). ------------------------------ Date: 19 Nov 1981 07:40:13-PST From: cbosgd!mark at Berkeley Subject: Re: systems news articles Having a newsgroup to discuss nets is different than discussing mail. I propose net.net and net.mail. I'm not sure net.net is needed - does fa.tcp-ip subsume it? There will probably soon be a net.csnet, too. Mark ------------------------------ From: Mike Muuss To: chico!duke!unc!smb at Ucb-C70 Subject: Networking Discussions in TCP-IP Digest? Steve - While the masthead "TCP-IP Digest" is really rather specialized, I had intended the Digest as more of a discussion on IMPLEMENTATION issues of networking (as opposed to Philosophical discussions as get found in HUMAN-NETS). The troubles with multiple networks, and the variety of message formats (for mail), and routing problems in general are all fair game for the TCP-IP digest. You are welcome to have this networking discussion in the TCP digest -- if the volume becomes too great I would be willing to clone a new digest later on. BRL polls Duke via UUCP, so messages addressed to ...!duke!bmd70!tcp-ip should make it to the digest (no need to go through Berkeley). Give it a try. Our RMAIL is smart enough to prevent accidental gatewaying; sorry. Cheers, -Mike ------------------------------ Sender: DRDAR-MST at OFFICE-8 Subject: TCP/IP and the CYBERS From: Richard Sitnik In a recent message Mike Muuss mentioned that there are plans to install a version of TCP-IP on the CYBER. Also, in Issue 6 of the TCP-IP Digest Mike mentioned that Tektronix has implemented TCP under NOS. I would like to find out more about plans regarding TCP-IP. In particular, do you know of anyone who is working on a NOS/BE version? I have spoken to Tektronix and their implementation will only work with NOS and requires a HYPER channel. They are not aware of any NOS/BE efforts. I will prepare an item for the TCP-IP Digest detailing my discussion with Tektronix. RICH END OF TCP-IP DIGEST ******************** ----MESSAGE-END----