ggm@brolga.cc.uq.oz.au (George Michaelson) (06/04/91)
jim@cs.strath.ac.uk (Jim Reid) writes: [cross-posted to prime the flame war again...] >In article <1367@exua.exeter.ac.uk> SAMcinty@exua.exeter.ac.uk (Scott McIntyre) writes: >Resistance to change is a factor, but not a major one. What is more >significant is that there is little expertise in the UK with running a >national IP based network and it is the lack of suitably qualified and >experienced people that makes things more awkward than they already are. Yes and then again no Jim. [who me? equivocate?] There was no experience of running a national IP network in Australia and substantial experience with the VMS JANET code. From this base, a national MULTI-PROTOCOL network but primarily IP based was established in under 1 year, and runs very very successfully. I'm sorry to disagree with you publically Jim, but doing TCP/IP is actually almost completely trivial. Nobody is obliged (or actually should) connect to a backbone until they are compliant with the standards (have a valid IP subnet, understand the DNS and routing) but that doesn't prevent the backbone being created in a VERY short space of time. Having done that, growth is pretty simple. Based on my experience (pre 1987) IP introduction in JANET is almost completely blocked by politics and bullsh. Technical issues are smokescreens. The truth in your statement is that there is woeful ignorance in UK campii about IP networking point blank. Its not ignorance of running a NATIONAL IP network, its ignorance of even running CAMPUS IP networks. We did it here in under a year. Its run by a 2-man team in Canberra and the goodwill of computer centres across the nation. Off-the-shelf technology, code built into virtually ALL *nix, available for VMS, VM, No mess, no fuss, no worries mate! -George -- George Michaelson G.Michaelson@cc.uq.oz.au The Prentice Centre | There's no market for University of Queensland | hippos in Philadelphia Phone: +61 7 365 4079 QLD Australia 4072 | -Bertold Brecht
tjc@ecs.soton.ac.uk (Tim Chown) (06/04/91)
In <1991Jun3.235516.7634@brolga.cc.uq.oz.au> ggm@brolga.cc.uq.oz.au (George Michaelson) writes: >jim@cs.strath.ac.uk (Jim Reid) writes: >>In article <1367@exua.exeter.ac.uk> SAMcinty@exua.exeter.ac.uk (Scott McIntyre) writes: >>Resistance to change is a factor, but not a major one. What is more >>significant is that there is little expertise in the UK with running a >>national IP based network and it is the lack of suitably qualified and >>experienced people that makes things more awkward than they already are. >Based on my experience (pre 1987) IP introduction in JANET is almost >completely blocked by politics and bullsh. Technical issues are >smokescreens. The truth in your statement is that there is woeful ignorance >in UK campii about IP networking point blank. Its not ignorance of running >a NATIONAL IP network, its ignorance of even running CAMPUS IP networks. >We did it here in under a year. Its run by a 2-man team in Canberra and >the goodwill of computer centres across the nation. Off-the-shelf technology, >code built into virtually ALL *nix, available for VMS, VM, No mess, no >fuss, no worries mate! Here at Southampton our Electronics and Computer Science departmental connections are all TCP/IP; we have 200+ nodes on our local network. Now we're just about to have our campus network go live, and yes, it all runs by Pink Book. Our Computing Services who will adminster this beauty have had it imposed on them from above by some national committee. Yum. Somewhere someone needs their head removing from their a**e !!! Tim --
ggm@brolga.cc.uq.oz.au (George Michaelson) (06/05/91)
Having posted, I decided I was being overly rude.
I went out and did some snmp netstats on the results of traceroute and
other goodies. Suffice to say that a sweeping statement about the IP
capabilities of the UK universities as a whole is a gross
overgeneralization, and there are quite clearly (from the routing
matrices returned by the ciscos of the UK) people out there in JANET
who know *exactly* what they are doing.
My comments about politics were scarcely more useful. I think they are true,
but probably not worth saying.
I'm interested in why layering over X.25 is being used, rather than MUXing
out bandwidth and using HDLC cards in the routers. would stealing 48k or 64k
splits from the X.25 lines really be that visible? The extra protocol
overheads of X.25-IP barely seem worth the cost savings to me. I'd rather
have the hardware anyday!
-To the networks people of the JANET TCP/IP group, I abjectly apologize
and take my (metaphorical) akubra off to you. I look forward to talk
sessions with family & friends over these links in the not-too-distant
future!
-George
--
George Michaelson
G.Michaelson@cc.uq.oz.au The Prentice Centre | There's no market for
University of Queensland | hippos in Philadelphia
Phone: +61 7 365 4079 QLD Australia 4072 | -Bertold Brechtzzassgl@uts.mcc.ac.uk (Geoff Lane) (06/05/91)
In article <7957@ecs.soton.ac.uk> tjc@ecs.soton.ac.uk (Tim Chown) writes: >Here at Southampton our Electronics and Computer Science departmental >connections are all TCP/IP; we have 200+ nodes on our local network. >Now we're just about to have our campus network go live, and yes, it all >runs by Pink Book. Our Computing Services who will adminster this beauty >have had it imposed on them from above by some national committee. Yum. >Somewhere someone needs their head removing from their a**e !!! > Tim Aha! Another fan of the famous JNT! The glorious committee which insists that everyone uses Janet protocols and only now are being dragged kicking and screaming into the mid 1970's. We here at Manchester also have hundreds of TCP/IP equiped PCs and workstations. We run our own name servers etc but as soon as we want to talk to some other site its back to the good old Janet protocols across X25 lines. If it were cheaper, or quicker or more reliable then it would make sense - but it is not. In its time the Coloured book protocols were a worthwhile attempt to bring some kind of organisation to computer networking when there were few if any standards, but the world has rejected these ideas and gone for something different. It would make some sense if the Coloured Book protocol software was freely available to anyone on request but it isn't. I attempted to get hold of the Pink Book s/w in order to port it to a new Unix machine we installed. It appears that I have to ask Amdahl, the suppliers of the hardware and OS, to place a contract with the S/W developers to port Pinkbook to the UTS system (a version of Sys V Unix) and then I have to ask the JNT to provide funds so we can then buy the ported s/w from Amdahl! Soon the agony will be over. By October (or sooner) there will be no physical or technical reasons why the UK should not be fully connected to the Internet. If non-UK sites still cannot finger at least some UK based systems then it will be due to politics pure and simple. This flamette was brought to you by me the undersigned and only represents the views of the users - not the administration - of this centre ;-) -- Geoff. Lane. Janet: zzassgl@uk.ac.mcc.cms UTS Sys Admin, Manchester Computing Centre, Oxford Rd, Manchester, M13 9PL
grahamt@syma.sussex.ac.uk (Graham S Thomas) (06/05/91)
From article <7957@ecs.soton.ac.uk>, by tjc@ecs.soton.ac.uk (Tim Chown): > > Here at Southampton our Electronics and Computer Science departmental > connections are all TCP/IP; we have 200+ nodes on our local network. > Now we're just about to have our campus network go live, and yes, it all > runs by Pink Book. Our Computing Services who will adminster this beauty > have had it imposed on them from above by some national committee. Yum. > Somewhere someone needs their head removing from their a**e !!! When the Coloured Book software suite was conceived, it was (according to my best information - correct me if I'm wrong) by no means clear that TCP/IP would become as dominant as it has today. In any case, the Coloured Books were always seen as ultimately giving way to full OSI protocols. Again, the originators could not have forseen that OSI would be such a long time a-comin', or that it could conceivably be threatened by semi-open protocols like TCP/IP. Admittedly, it might have taken a little too long for the JANET management to face up to commercial realities (i.e. TCP/IP comes cheaper and quicker for new machines) and user desires (the rest of the world - exaggerating a bit - uses it, so why can't we?). But early this year the decision was taken to offer TCP/IP over JANET as a supplementary protocol. My impression at this April's JANET Networkshop was that there is a lot of effort being put into its implementation, and there are a lot of potential users keen to see that this effort doesn't flag. So, hang on in there until (maybe) the beginning of '92. Already, you can transfer files to & from the Internet using uk.ac.ft-relay, and it's possible to get an account on uk.ac.nsfnet-relay from which you can telnet across the world. So it's not all doom and gloom! Graham -- Graham Thomas, SPRU, Mantell Building, U of Sussex, Brighton, BN1 9RF, UK Email: grahamt@syma.sussex.ac.uk Phone: +44 273 678165 Fax: .. 685865
ccrth@lut.ac.uk (Rob Thirlby) (06/06/91)
Keywords: The decision to use IP tunelling rather than multiplexing our access lines to janet reflects the very variable and unpredictable demand from sites and the wide disparity of bandwidth of the access lines. Also many of us are hoping to use tunelling facilities in our existing kit such as netcomm switches/ECB rather than buy unnecessary routers. We are getting there. The pilot scheme is being persued with vigour and many sites such as mine will ensure that the pressure to offer a service is kept up! After all isnt it said in religious circles that late converts make the most devout members? Yrs, Rob Thirlby
J.Crowcroft@CS.UCL.AC.UK (Jon Crowcroft) (06/06/91)
>I'm interested in why layering over X.25 is being used, rather than MUXing >out bandwidth and using HDLC cards in the routers. would stealing 48k or 64k >splits from the X.25 lines really be that visible? The extra protocol >overheads of X.25-IP barely seem worth the cost savings to me. I'd rather geo: To split out bandwidth by muxing would be to pre-decide how much... and would cost some for muxes... the current choice represents expediency and manageablity...(though for more money, we could buy snazzy controllable muxes). the routers have 2Mbps lines to the backbone x.25 switches - we can tune the amount shared twixt native x.25 and IP ... by varying clocks etc... for well engineered (read giant window & packet size) x.25 at 2Mbps, you dont get x.25 window blocking IP, so minimal overhead and neet enforcement of share... experience in other nets of IP on x.25 usually had poor x.25 implementations (read window 2, max packet size 128), and so negative comments from elsewhere are often (not always) discounted (though less by me:-). of course, I would prefer frame relay to X.25 full protocol, then there wouldnt be weird interactions between retransmits at x.25 and above IP, but we await this...in any case, since the backbone cable plant has very low error rate, we see this rarely...more often is interaction twixt congestion control inside the switch cloud, and congestion avoidance twixt TCP end points... Another path to IP involves the x.25 backbone switches running the IP tunnel (and later migrating to run this over frame relay internally) this is being tried/evaluated...as will be isdn & SMDS... jon p.s. if you run an AFS filesystem that we can access, please let us know - we are currently seeing the usual set of places (transarc/cmu/mit etc etc) but would find it amusing to go even further afield like as far as OZ:-)
PADLIPSKY@A.ISI.EDU (Michael Padlipsky) (06/07/91)
Graham Thomas-- I'd be so fascinated to see an explanation of your depiction of TCP/IP as "SEMI-open" [emphasis added] that I'd even forego the pleasure of debating whether the originators of the "Coloured Books" SHOULD have realized [or, more locally appropriate, realised] that OSI would be a very long time in coming, in exchange. cheers, map -------
ggm@brolga.cc.uq.oz.au (George Michaelson) (06/07/91)
I suspect that my lack of knowledge of the internal architecture of the IP/X25 connections is the source of some confusion for me. -I'm assuming that you are doing IP tunnels to one common endpoint which is feeding into the Internet, and thus there is no "logical" common backbone beyond a star-hub (a cisco in ulcc or ucl?). I'm also assuming that down the track this will not scale well, and some degree of distribution of the load will prove neccessary. -That means that IP routing boxes will be inter-connecting to form a backbone, and there WILL be some kind of shared usage of trunk lines. Question: would *THAT* traffic be done using X.25 tunnels as well? based on those assumptions you can now go and destroy all of what follows! > ccrth@lut.ac.uk (Rob Thirlby) writes: >The decision to use IP tunelling rather than multiplexing our access >lines to janet reflects the very variable and unpredictable demand from >sites and the wide disparity of bandwidth of the access lines. Also >many of us are hoping to use tunelling facilities in our existing kit >such as netcomm switches/ECB rather than buy unnecessary routers. I have some doubt thats going to prove "optimal" in the longer term, but I also wonder if how individual campii connect into a backbone should dictate how the backbone allocates its bandwidth? I wouldn't want the NSF to allocate 48k trunks simply because AARN is only using 48k for the drops! I think this depends on what you think the long-term implications of a wide-band connectionless network are. What sort of dataflow down the shared links do you expect? If you tunnel each campus connection over X.25, will some pool of JANET backbone X.25 routers do the right thing with all the packets going to the US and onward? I'd expect that the use of long-haul links will look very much the same for you as they do for everybody else: NNTP, SMTP and FTP traffic will be semi-continuous, with a layering of telnet and other small packet traffic during "normal working hours". -That means that shared resources will be blasted with big packets all flowing in pretty much the same direction: outer-lying places to FTP archives outer-lying places to newsfeeds outer-lying places to mailhubs everywhere to the International fatpipes. Surely having boxes which understand the IP addressing and co-elesce all this into more optimal flow down HDLC is better than maintaining 1001 separate X.25 tunnels all going to the same place? >We are getting there. The pilot scheme is being persued with vigour and >many sites such as mine will ensure that the pressure to offer a >service is kept up! After all isnt it said in religious circles >that late converts make the most devout members? Point taken. I must lay claim to fitting that pattern, I doubt if anybody in the UK would recall me being exactly the most pro-TCP of people. However like the small boy in the "Bubbles" advert: since using their soap, I have used no other. I really hope the project continues to flourish. If I'm provoking any anger please accept my apologies, it's a reflection of the long long time this has been coming. -Roll on '92! -George -- George Michaelson G.Michaelson@cc.uq.oz.au The Prentice Centre | There's no market for University of Queensland | hippos in Philadelphia Phone: +61 7 365 4079 QLD Australia 4072 | -Bertold Brecht
Andy.Linton@comp.vuw.ac.nz (Andy Linton) (06/07/91)
In article <1991Jun3.235516.7634@brolga.cc.uq.oz.au>, ggm@brolga.cc.uq.oz.au (George Michaelson) writes: |> Based on my experience (pre 1987) IP introduction in JANET is almost |> completely blocked by politics and bullsh. Technical issues are |> smokescreens. The truth in your statement is that there is woeful ignorance |> in UK campii about IP networking point blank. Its not ignorance of running |> a NATIONAL IP network, its ignorance of even running CAMPUS IP networks. |> My experience in the UK is a little more recent than George - I left in mid 89 but I'd support his stance on this. The politics of the JNT with regard to equipment purchases prevented people buying decent router boxes to implement campus IP networks since they weren't "Protocol Police approved". The network in Newcastle and Durham has changed since I left but it wasn't ready then to be connected into the Internet. |> We did it here in under a year. Its run by a 2-man team in Canberra and |> the goodwill of computer centres across the nation. Off-the-shelf |> technology, |> code built into virtually ALL *nix, available for VMS, VM, No mess, |> no |> fuss, no worries mate! Snap, both our networks are smaller than the UK set up - we had things running in a similar time scale, we don't have any full time staff as yet and rely also on the goodwill of university computer centres and government departments across the country. I look forward to seeing the UK connected as I sure George does. I note with interest the comment of one of my UK colleagues last week - I won't name him - you never know who's listening (:-). Talking about the experimental IP links in the UK, he said: "With this sort of progress with luck we will be able to avoid falling yet further behind the US networkwise!"
pcg@aber.ac.uk (Piercarlo Grandi) (06/15/91)
On 5 Jun 91 13:39:32 GMT, grahamt@syma.sussex.ac.uk (Graham S Thomas) said: grahamt> When the Coloured Book software suite was conceived, it was grahamt> (according to my best information - correct me if I'm wrong) by grahamt> no means clear that TCP/IP would become as dominant as it has grahamt> today. On this I would disagree. At the time the CB were done, the *only* large scale network, especially in the academic sector, was the ARPAnet. There were *no* ISO style networks in widespread use, in the academic sector. Also, the DARPA was funding a lot of efforts to produce networking ARPAnet software for a large variety of machines, software that was, thanks to US govmt. practice, freely available. The JNT chose unproven, incomplete technology for which there was no ready made software, against proven, well established technology for which there was a large and growing mass of ready made software. This was clear even without the benefit of hindsight. Anybody who used the ARPAnet knew that at its time it was simply unique, it did work well, and its technology had several years of advantage on anything else. grahamt> In any case, the Coloured Books were always seen as ultimately grahamt> giving way to full OSI protocols. Again, the originators could grahamt> not have forseen that OSI would be such a long time a-comin', You mean that somebody at the time did not realize that an incomplete set of protocols still being actively being discussed and designed was not 'a long way in coming'. I find this astonishing. Especially as ARPAnet protocols were not only already discussed and designed, they were also already implemented. It is a well known fact that international standadization is already a slow process when the standard sanctifies existing technology, and delays become even more dramatic when the technology has to be developed by the standards bodies themselves. There is also the obvious political observation that, just like ISDN, monopolistic PTTs, especially in Europe, have no interest in rushing new technology to the market, because it means throwing away a lot of existing investment, when there is no need, because the market has to bear what the PTT decides. X.25 networking has been boycotted with great zeal by all/most European PTTs, just like any form of data traffic (which has been boycotted in the USA by AT&T, as Tanenbaum notes), and this was very well known among observers of the telecom scene. Your information, IMNHO, seems to imply that the originators of JANET were naive idealists with thoroughly unrealistic optimistim on the maturity and development speed of ISO technology. Amusing :-(. grahamt> or that it could conceivably be threatened by semi-open grahamt> protocols like TCP/IP. Ah, now we understand. By 'semi-open' you are possibly implying that the ARPAnet suite of procols had the fatal defect of being designed under the auspices of and controlled by somebody else than the cosy club of European PTTs, which could slow down its development at will, but by a relatively efficient Agency of the US Govmt... After all JNT is something that emanates from the British Govmt., which at the time was also the owner of the local PTT. As they say "a synergetic partnership :-). -- Piercarlo Grandi | ARPA: pcg%uk.ac.aber@nsfnet-relay.ac.uk Dept of CS, UCW Aberystwyth | UUCP: ...!mcsun!ukc!aber-cs!pcg Penglais, Aberystwyth SY23 3BZ, UK | INET: pcg@aber.ac.uk
igb@fulcrum.bt.co.uk (Ian G Batten) (06/17/91)
In article <PCG.91Jun14183028@aberdb.aber.ac.uk> pcg@aber.ac.uk (Piercarlo Grandi) writes: > On 5 Jun 91 13:39:32 GMT, grahamt@syma.sussex.ac.uk (Graham S Thomas) said: > > grahamt> When the Coloured Book software suite was conceived, it was > grahamt> (according to my best information - correct me if I'm wrong) by > grahamt> no means clear that TCP/IP would become as dominant as it has > grahamt> today. > > On this I would disagree. At the time the CB were done, the *only* large > scale network, especially in the academic sector, was the ARPAnet. There > were *no* ISO style networks in widespread use, in the academic sector. But were there TCP networks either? My understanding is that at the time the CBen were being written, the majority of the ARPAnet was running NCP. TCP was certainly not the proven technology that it is now. Remember that TCP was first seriously available in 4.2bsd. I don't have an easy feel for when ``most'' Universities would have had something running 4.2bsd, but I feel certain that the number of Computer Centres who would have traded their big iron for a 780 running 4.2bsd was vanishingly small. That may be right or wrong, but it is a fact. Additionally the PTTs, especially BT, were installing shed-loads of 8 bit X25. IP over X25 was inefficient to put it mildly and without a massive Military-Industrial complex (TM) there was no way that a datagram infrastructure was going to be installed. (**) (*) Did you know the post office still has a class A IP number from the early days of EPSS? (**) Unless you had voted SERC out and taken the MOD instead. Remember who funded most of US Computer Science through that period? ian
exspes@gdr.bath.ac.uk (P E Smee) (06/17/91)
In article <PCG.91Jun14183028@aberdb.aber.ac.uk> pcg@aber.ac.uk (Piercarlo Grandi) writes: >Also, the DARPA was funding a lot of efforts to produce networking >ARPAnet software for a large variety of machines, software that was, >thanks to US govmt. practice, freely available. But not always outside the States -- owing to various 'technology transfer' regulations, including CoCom. And, the fairly arbitrary nature of CoCom means that outside the US, you've got to be a bit leery of using things which were funded by the US military, since even if they are available today, they may not be tomorrow. >The JNT chose unproven, incomplete technology for which there was no >ready made software, against proven, well established technology for >which there was a large and growing mass of ready made software. Except, this 'large mass' of software has some serious deficiencies (as compared with notional IP specs -- particularly, in terms of subnet masking. It also contains many well-known security holes, which may not have been so important when only real ARPA-approved sites were netted, but becomes more so as the net becomes more open. (Anyone using any of the 'r' commands, or NFS write-mode mounts, is more or less asking to be hit, among other things.) Also, the underlying structure of addresses makes address management a royal pain. And, in some places (e.g. here) we do not have available, for cultural reasons, the large mass of student employees which a typical US University used to handle all this drudgery. (Mind you, the ISO NSAP addresses seem to have gone overboard in the other direction. -- Paul Smee, Computing Service, University of Bristol, Bristol BS8 1UD, UK P.Smee@bristol.ac.uk - ..!uunet!ukc!bsmail!p.smee - Tel +44 272 303132
pcg@aber.ac.uk (Piercarlo Grandi) (06/18/91)
On 17 Jun 91 11:50:11 GMT, exspes@gdr.bath.ac.uk (P E Smee) said:
exspes> In article <PCG.91Jun14183028@aberdb.aber.ac.uk> pcg@aber.ac.uk
exspes> (Piercarlo Grandi) writes:
pcg> Also, the DARPA was funding a lot of efforts to produce networking
pcg> ARPAnet software for a large variety of machines, software that was,
pcg> thanks to US govmt. practice, freely available.
exspes> But not always outside the States -- owing to various 'technology
exspes> transfer' regulations, including CoCom. [ ... even staunch
expses> allies like the UK have to be wary of crazy USA export controls
expses> ... ]
Agreed. But this does not mean then that JNT had to choose something
completely different and incompatible and build everything from the
ground up. It could have meant that JNT could have taken what was on
offer (a lot!) as to implementations and start from the publicly
available standards to duplicate what was not.
Frankly, to everybody but a PTT, the DARPA is a far more "open" body
than the CCITT. For example anbybody can submit cheaply and have a quick
"approval" for an RFC, if it is useful. No red tape, no colossal
expenses.
pcg> The JNT chose unproven, incomplete technology for which there was
pcg> no ready made software, against proven, well established technology
pcg> for which there was a large and growing mass of ready made
pcg> software.
exspes> Except, this 'large mass' of software has some serious deficiencies
I would object to 'serious', it is a classic english understatement; I
would say 'catastrophic'.
There are not that many major sw/hw components in widespread use today
which do not have *catastrophic* shortcomings. SunOS, SysV3.2, MSDOS,
NFS, X11, ... choose your favourite abomination (not to mention ASN.1,
X.400, the plethora of ridiculously incompatible TPs, ...).
TCP/IP and company, for all their failings (my favourite one is
watermark acknowledgement in TCP/IP) and limitations (no really good
internetwork gateway or routing protocols) is not one of the worst, and
it is lightweight and efficient, at least compared to ISO/OSI.
exspes> (as compared with notional IP specs -- particularly, in terms of
exspes> subnet masking.
The problem with subnet masking is that it is basically a hack; it could
well be a clean hack, but very few people actually understand
subnetting. Too bad.
exspes> It also contains many well-known security holes, [ ... ] Also,
exspes> the underlying structure of addresses makes address management a
exspes> royal pain.
Just flies in the ointment. The comparison the JNT had before themselves
was between a working viable fairly suboptimal system of specifications
and implementations and an incomplete paper design, which could have
been conceivably optimally designed and implemented (it wasn't...), but
nobody could have been sure.
*IF* ISO/OSI had been completely designed then, *IF* the design had been
problem free, *IF* implementations had been freely available, then maybe
it could have made sense for the UK to ignore a dozen years of ARPAnet
experience and easy connectivity with the largest (and at the time the
only one) research network in the world and live in splendid isolation.
But all those *IF*s were not true at the time, only very prejudiced
people could not see it, and are still not true now, and even if they
were, that is even *IF* ISO/OSI were distinctly better than ARPAnet as
to technical quality and availability, it would still make more sense to
have connectivity, and live with good enough.
exspes> And, in some places (e.g. here) we do not have available, for
exspes> cultural reasons, the large mass of student employees which a
exspes> typical US University used to handle all this drudgery. (Mind
exspes> you, the ISO NSAP addresses seem to have gone overboard in the
exspes> other direction.
Here the culprit is not really TCP/IP, but UNIX. UNIX, which was
designed assumign that Thompson & Ritchie were the system managers.
TCP/IP is really a minor nuisance, and there are decent tools out there
to relieve a lot of the problems.
As I said we are stuck with abominable sw all around us; yet we still
muddle thru.
For a lot of this the UK academic environment is guilty of many sins of
omission; the quality of much UK sw research has been, especially in the
sixties and the seventies, much higher than that of now popular USA
equivalents, but the stubborn persistence with which the UK research has
been kept well "hidden" has been a disgrace. Just compare MUSS with
Unix...
--
Piercarlo Grandi | ARPA: pcg%uk.ac.aber@nsfnet-relay.ac.uk
Dept of CS, UCW Aberystwyth | UUCP: ...!mcsun!ukc!aber-cs!pcg
Penglais, Aberystwyth SY23 3BZ, UK | INET: pcg@aber.ac.ukadam@castle.ed.ac.uk (Adam Hamilton) (06/18/91)
In article <PCG.91Jun17234451@aberdb.aber.ac.uk> pcg@aber.ac.uk (Piercarlo Grandi) writes:
: Agreed. But this does not mean then that JNT had to choose something
: completely different and incompatible and build everything from the
: ground up. It could have meant that JNT could have taken what was on
: offer (a lot!) as to implementations and start from the publicly
: available standards to duplicate what was not.
:
: The JNT chose unproven, incomplete technology for which there was
: no ready made software, against proven, well established technology
: for which there was a large and growing mass of ready made
: software.
:
: Just flies in the ointment. The comparison the JNT had before themselves
: was between a working viable fairly suboptimal system of specifications
: and implementations and an incomplete paper design, which could have
: been conceivably optimally designed and implemented (it wasn't...), but
: nobody could have been sure.
:
: *IF* ISO/OSI had been completely designed then, *IF* the design had been
: problem free, *IF* implementations had been freely available, then maybe
: it could have made sense for the UK to ignore a dozen years of ARPAnet
: experience and easy connectivity with the largest (and at the time the
: only one) research network in the world and live in splendid isolation.
Most of this is simply wrong. The choice made by the JNT
was NOT between OSI-on-paper and a basically-working-TCP-IP. It was between
TCP-IP-on-paper and a basically-working-X25, the only standardised available
protocol of the time. The decision to target on OSI when available is
hardly one to be criticised; after all, the Internet has stated it will
convert eventually.
PCG seems to have the timescales all wrong. As has been
explained to him several times, the Arpanet was NOT running well-proven
software at the time, it did NOT have a dozen years of experience and
it did NOT have a "working viable...system of specifications and
implementations". The key point he misses is that what makes sense is to
run the same protocols locally as you will have to connect to if you want
to go further afield. In Europe at that time that meant X25.
Next message, he will start to tell us again about how the JNT chose
big-endian domain ordering when little-endian had already been chosen for
TCP/IP (and for viewers who have missed this endless discussion - they didn't.
They simply failed to change their minds some months later when the
opposite decision was made over the pond).pcg@aber.ac.uk (Piercarlo Grandi) (06/18/91)
On 17 Jun 91 07:32:13 GMT, igb@fulcrum.bt.co.uk (Ian G Batten) said:
igb> In article <PCG.91Jun14183028@aberdb.aber.ac.uk> pcg@aber.ac.uk
igb> (Piercarlo Grandi) writes:
pcg> On 5 Jun 91 13:39:32 GMT, grahamt@syma.sussex.ac.uk (Graham S Thomas) said:
grahamt> When the Coloured Book software suite was conceived, it was
grahamt> (according to my best information - correct me if I'm wrong) by
grahamt> no means clear that TCP/IP would become as dominant as it has
grahamt> today.
pcg> On this I would disagree. At the time the CB were done, the *only*
pcg> large scale network, especially in the academic sector, was the
pcg> ARPAnet. There were *no* ISO style networks in widespread use, in
pcg> the academic sector.
igb> But were there TCP networks either? My understanding is that at the
igb> time the CBen were being written, the majority of the ARPAnet was
igb> running NCP. TCP was certainly not the proven technology that it is
igb> now.
Yes, buit I was really referring to 'ARPAnet technology', quite
explicitly, rather than 'Internet technology'. ARPAnet technology was
indeed working, was proven, and was obviously viable, and with an
obviousn growth path, and was the only technology for WANs at the time.
igb> Remember that TCP was first seriously available in 4.2bsd. I don't
igb> have an easy feel for when ``most'' Universities would have had
igb> something running 4.2bsd, but I feel certain that the number of
igb> Computer Centres who would have traded their big iron for a 780
igb> running 4.2bsd was vanishingly small. That may be right or wrong,
igb> but it is a fact.
I agree on the fact, but maybe you miss out that ARPAnet was a Big Iron
type of network. CS departments with BSD minis were a very late
addition, and one that DARPA had to foster by financing the development
of TCP/IP for BSD. But the ARPAnet did not start with VAXes and BSDs;
initially the ARPAnet was a WAN to which dozens of IBM, Honeywell, DEC
mainframes in corporate and Unviersity computer centres were connected.
Cost of connection was high enough that few departments could afford it.
igb> Additionally the PTTs, especially BT, were installing shed-loads of
igb> 8 bit X25.
IMNHO they were try to soft pedal introducing X.25 as much as they
could, and to install the minimal X.25 services thatthey could get away
with, while paying very political lip service to the concept. For an
European PTT x.25 networks were a stupid nuisance, a distraction, and an
opportunity for computer businesses to corner the lucrative services
market while the PTT eked out small money from just carrying the bits
around.
As of now, advanced X.25 technology in Europe gives you 2400 baud
connectivity, just as one example, and international X.25 links are a
few *times* slower and more expensive than a TrailBlazer link. Technical
problems? No, simple ill will. The IP network that has sprouted recently
in Europe puts the PTT run X.25 networks to shame as to bandwidth
supported, cost of traffic, and the velocity with which it has been set
up. And please note that it still has to run on the incredibly
overpriced leased lines that European PTTs deign to offer customers. In
Europe you don't get any cheap T1/T3 style links, nor you can set up
your own microwave network that easily, unless you are a very big
corporation or the military.
igb> IP over X25 was inefficient to put it mildly
Well, one could have taken ARPAnet technology and run their protocols
lock stock and barrel. What we have now *is* indeed IP over X.25, ten
years late, even it is not that inefficient, if done cleverly.
igb> and without a massive Military-Industrial complex (TM) there was no
igb> way that a datagram infrastructure was going to be installed. (**)
The UK does have a military industrial complex, and a relatively big one
(consider the money spent on military research), but one that is very
closed and inward looking. The USA have been luckier in this respect.
For example, whatever the nasty militaristic and corporatist overtones
of the USA military industrial complex, it funds 60-80% of research at
major USA Universities, for example MIT, and in a fairly open way.
Unheard of in the UK.
igb> (*) Did you know the post office still has a class A IP number from the
igb> early days of EPSS?
I did not, actually. Amusing little bit of news.
igb> (**) Unless you had voted SERC out and taken the MOD instead. Remember
igb> who funded most of US Computer Science through that period?
This is what has happened, I understand. Reading between the lines it is
fairly clear that the military have been leaning quite heavily on the
folks at education and industry to get better connectivity to their
chummy colleagues across the Pond.
Let me guess that since the nice chaps with the Stars and Stripes have
very effective intelligence networks based on Internet technology, the
UK military have found it very useful (Falklands, Kuwait) to be albe to
connect into those.
--
Piercarlo Grandi | ARPA: pcg%uk.ac.aber@nsfnet-relay.ac.uk
Dept of CS, UCW Aberystwyth | UUCP: ...!mcsun!ukc!aber-cs!pcg
Penglais, Aberystwyth SY23 3BZ, UK | INET: pcg@aber.ac.ukggm@brolga.cc.uq.oz.au (George Michaelson) (06/19/91)
adam@castle.ed.ac.uk (Adam Hamilton) writes: > Next message, he will start to tell us again about how the JNT chose >big-endian domain ordering when little-endian had already been chosen for >TCP/IP (and for viewers who have missed this endless discussion - they didn't. >They simply failed to change their minds some months later when the >opposite decision was made over the pond). There are more versions of this story than I care to think about. I see no reason to believe yours over anybody elses. I suggest somebody who was present at the JNT sponsored meeting(s) of the time should comment. Were you? I believe Jim Cragie, John Larmouth or Steve Kille are all qualified to comment (I doubt if they will by choice, this opens old and ugly wounds). I'm not saying you're wrong, I'm just asking that you offer some proof of correctness for your version. -george -- George Michaelson G.Michaelson@cc.uq.oz.au The Prentice Centre | There's no market for University of Queensland | hippos in Philadelphia Phone: +61 7 365 4079 QLD Australia 4072 | -Bertold Brecht
pcg@aber.ac.uk (Piercarlo Grandi) (06/20/91)
On 18 Jun 91 08:40:31 GMT, adam@castle.ed.ac.uk (Adam Hamilton) said:
adam> In article <PCG.91Jun17234451@aberdb.aber.ac.uk> pcg@aber.ac.uk
adam> (Piercarlo Grandi) writes:
adam> The choice made by the JNT was NOT between OSI-on-paper and a
adam> basically-working-TCP-IP. It was between TCP-IP-on-paper and a
adam> basically-working-X25, the only standardised available protocol of
adam> the time.
No, it was between working ARPAnet with many services and dozens of
nodes and X.25 yet to be used massively and little more than remote
login.
The choice between Internet and ISO/OSI is the one they are facing now,
or rather, the choice has already been made by somebody else, and the
JNT are trying to ride the wave.
adam> The decision to target on OSI when available is hardly one to be
adam> criticised; after all, the Internet has stated it will convert
adam> eventually.
Milnet, maybe will (in due time :->). NFSnet, quite improbably. The
Internet, surely not.
As to 'OSI when available' I would never bet the farm on a set of
standards yet to be approved, sight unseen. Apparently I am in good
company.
adam> the Arpanet was NOT running well-proven software at the time, it
adam> did NOT have a dozen years of experience and it did NOT have a
adam> "working viable...system of specifications and implementations".
Tell that to the guys that were running the ARPAnet at the time... :-)
Maybe I suffer from hallucinations, as somebody else has pointed out
before. I have obviously delusionary memories of using the ARPAnet in
the late seventies/early eighties with remote login, mail, file
transfer between dozens of otherwise incompatible mainframes.
I cannot remember any similarly complete ready made alternative from the
X.25/ISO/OSI camp.
Still I must admit that the only way I had to get onto ARPAnet from
Italy was to use TYMnet, which was X.25 based. But X.25 based networks
did not offer much more than remote login, and painfully as to that.
adam> The key point he misses is that what makes sense is to run the
adam> same protocols locally as you will have to connect to if you want
adam> to go further afield. In Europe at that time that meant X25.
Mention *any* other largish X.25 based research network in Europe at the
time. Or even now.
Or maybe this was another JNT miscalculation. They may well have
believed that there would be a lot of traffic with other European
countries and with Australia and New Zealand, rather than with the USA.
Wishful thinking, if so. A dozen years ago I was using TYMnet to connect
to the ARPAnet, not to CYGALE or EIN. For good reasons...
adam> Next message, he will start to tell us again about how the JNT
adam> chose big-endian domain ordering when little-endian had already
adam> been chosen for TCP/IP (and for viewers who have missed this
adam> endless discussion - they didn't. They simply failed to change
adam> their minds some months later when the opposite decision was made
adam> over the pond).
This is not what I wrote. I remember (and will eventually find the
article I saved) one of the guys who did attend the meetings reported
that JNT had chosen a completely different addressing format with big
endian domains, and then they changed their minds and decided to use the
USA adopted syntax but with their own original big endian domain
ordering. Hearsay, admittedly. Will try to retrieve that posting.
--
Piercarlo Grandi | ARPA: pcg%uk.ac.aber@nsfnet-relay.ac.uk
Dept of CS, UCW Aberystwyth | UUCP: ...!mcsun!ukc!aber-cs!pcg
Penglais, Aberystwyth SY23 3BZ, UK | INET: pcg@aber.ac.uklaws@CCINT1.RSRE.MOD.UK ("John") (06/21/91)
Ian, I agree with your history. It does seem to agree with my memory of events. For the last 12 years I have been responsible for funding and the use of the UK MOD interconnection to the US Internet (nee Arpanet). During that period and prior to it I have had involvement with the JNT. BTW for many years I carried academic traffic at my cost; more recently we (I & JNT) have agreed to joint funding of the link. I still have an early 80's lapel button "I survived the TCP transition" and I can recall C-Book discussions well before that date. I also recall that I was using BT (then GPO) X.25 in the late 70's. It is some regret, to me at least, that my Country no longer knows how to judge the time and place to invest in a Grand Challenge (probably has something to do with the accountants moving in - value for money). In the late 60's and early 70's the UK was co-equal with the US in networking. There was also an internetworking (EIN) programme in Europe throughout the 70's (see early ICCC Proceedings). The JNT is responsible through various academic boards to the Dept of Education & Science. In the mid/late 70's it would be unlikely to have a winning case to the Boards for networking based upon a still developing concept that was largely funded by a Defense community; much better to have a case based upon concepts being developed in the civil/vendor community. John
mickey@bigglescetia.dml (M J Dance Account) (06/21/91)
Well, looks like I may well unsubscribe to this group as for the past few weeks the only traffic I have seen is about Janet and ARPA all of which has little to do with british culture (well I suppose that Janet *WAS* set up by british people but hardly a majority of them). ANyway, it seems that people are haggling over time-scales so I'll add my comments. 1) I do not known the exact time scales *BUT* I was doing research in an English university in the period 1984-1987 and Janet was pretty new at that time (althoug it did exist). ARPA net also existed as my research weas in large networks and there were *NO* large british networks so most of the papers I read were from the USA, an awful lot were from BBN on the implementation of routing in the ARPA net. 2) At this time most published material (which of course lags behind installed systems) was about IMPs connecting diverse machines and the word UNIX was almost never seen. 3) The coloured book protocols (Janet) existed but I never saw a working installation (my university was not a very large one) although major (in size) universities were inter-connected. 4) Almost all local network research was based on Basic Block Protocol on Cambridge rings (a sort of token-ring network) which as far as I am aware never left England (well it reached Scotland:)) 5) As a result of this all that I known about TCP/IP I have learnt since (and this learning was more or less obligatory given that I work for a computer constructor that sells work-stations). 6) Perhaps things have changed but all=most all that I learnt about networking protocols (with the exception of routing protocols) during my research has been of no use to me at all. IMHO this is why England lags behind in networking (well that and British Telecoms pricing strategy) For those who are persistent enough to get this far, what are you hoping to achieve. Thos who think Janet is wonderful are not going to change their minds and vv those who think the opposite are unlikely to change their minds. I doubt that those who decide British university policy ever read this group (I even doubt that most of them even use a network:).