Welcome to the third hypertext issue of news@UK.


Volume 3 Number 6 (December 1994)

Editor's Column

(Susan Small)

Welcome to the last issue of news@UK for 1994. As the newsletter is now three years old, it seems like a good time to take a step back and ask for your views on the contents and format. You should therefore find a questionnaire insert which I hope you will take the time to complete and return to Owles Hall. If you don't like filling in questionnaires, you could always let us have your views in a letter, or via e-mail.

My thanks must go to all those people who have contributed over the year to making news@UK an interesting read.

We are always looking for reviewers of various products, including books and software. If you are interested in writing a review of a new product, please let me know and we will ask for a review copy on your behalf. Alternatively if there is something that you think we should review, please let me know and we will put the review out to tender!

Season's Greetings and a Prosperous 1995.



Around Europe

Across the Pond

From the Net



Report from the Chair

(Mick Farmer)

Well, my request for feedback from you out there in the last issue certainly made some of you stir! Your editor received a number of letters (a rare event indeed), two of which are published elsewhere in this issue. The main topic of discussion was certainly the new Linux SIG that we intend to start very soon. This is a departure for the UKUUG, in the sense that we intend to offer membership to non-members of the UKUUG (members of the UKUUG are automatically members of any SIG they choose) for an annual fee. It has been a long process deciding exactly what services are offered to those joining only the Linux SIG. It now looks that an annual subscription of about 20 GB pounds will give our new recruits a regular Linux newsletter (all UKUUG members will receive this automatically) and access to Linux-related material at UKUUG membership prices; this includes books and workshops. This package will attract many people, some of whom (I hope) will see the advantage in full UKUUG membership, and join the UKUUG in order to contribute on a wider front than just Linux. Follow the links to our Linux SIG WWW page for the latest details. There is also an e-mail distribution list devoted to this SIG, send e-mail to [email protected] with a message body subscribing to linux-uk-usergroup to join the discussions on the formation of our Linux SIG.

As most of you know, the Sun UK User Group (SUKUG) also uses Owles Hall for its secretariat. The SUKUG have recently installed a new Sun Sparc system at Owles Hall and the UKUUG are sharing the system with them (as are EurOpen). The UKUUG is contributing by providing network access and possibly by providing monies for larger disks. Our ultimate aim is to move the UKUUG Web pages to this system, and eventually provide various membership access facilities (thank you SUKUG).

I attended the recent EurOpen Governing Board meeting in Bucharest (see Andrew Macpherson's report elsewhere in this issue) where a number of issues affecting the UKUUG were discussed. The most immediate was the announcement that the Irish UNIX User Group (IUUG) had folded for various reasons. It was agreed that we, the UKUUG, would offer any former member of the IUUG membership of the UKUUG until such time that the IUUG is re-constituted. I welcome any ex-IUUG member to our fold.

I also attended the recent ICT Round Table 8 meeting in Brussels (see report elsewhere in this issue). In my capacity as chair of the UKUUG, we have become involved with a project providing information from the EC via WWW servers around Europe (provided by EurOpen). I hope to give you fuller details in the next newsletter.

Your Council has been busy recently planning new events for the coming year, most of which are still in the planning stage. However, we have Rik Farrow (ex- Technical Editor of Open Computing) giving a three day tutorial/course on Advanced UNIX and Internet Security on 8-10 February 1995. I am hoping that details will be finalised in time for an insertion in this issue – look for the coloured pages! In the pipeline are events concerning the World Wide Web (both servers and clients) and Tcl/Tk programming. We have asked acknowledged experts in these fields to lead these events and hope to have responses very soon. As always, consult our Web pages or the uk.ukuug newsgroup for the latest information.

I'd love to see an expansion in local groups, but my principal worry is that they may be too focused on subjects. If you go out for a regular drink with colleagues, why not think of setting up a LUG? The UKUUG will even help financially. Now that's a good offer!

Please take time to read the letters column. It's obvious to me that many of you care (almost passionately) about the future of the UKUUG, UNIX, and even Open Systems? I know I'd love to receive more letters or e-mail concerning our group. Come on, get involved!

Cambridge LUG

(Jane Shute)

Our last meeting was held on 20 October, kindly hosted by Olivetti Research.

The first speaker was Mike Peach from MicroMuse. He delivered an overview of the history, design and possible future of network management protocols, including SNMP and SNMPv2. He also mentioned the various software that is available to manage networks.

The second speaker was Tim Addison from the Cambridge Sun office. He introduced three new products from Sun. The first was the Netra Internet Server this consists of a Sparc 5, or Sparc 20, that is pre-configured for attaching to the Internet. It includes Access, Usenet, Mosaic, WAIS, Gopher and e-mail. There is also a Firewall option that is supported.

The second product is the Netra System Management server. This attempts to centralise the system management of PCs via a new product called Solarnet, in addition there is a backup solution available called Networker.

The final product is Solaris 2.4, it should be released around December or January, I believe. It includes Motif libraries which will allow you to compile X Windows programs. In addition it provides performance enhancements for smaller machines, e.g. those with single processors with less than 16 Mb of memory.

The next meeting is 1 December at the Department of Geography, Downing Place – kindly hosted by Dan Scott. Our presenters will be IXI talking about Eye2eye.

London LUG

(Andrew Findlay)

The view from Brunel

LUUG joined with the Institution of Electrical Engineers for a lecture on 24 November 1994. The talk was on video compression techniques for packet data networks, and the speakers were Ian Wakeman and Mark Handley of UCL. True to the topic, the lecture was conducted over the Internet: Ian was at Brunel University in Uxbridge with an IEE audience, and Mark was at UCL in central London with a LUUG audience.

Video and audio was relayed between the two using the SuperJANET data network. A variety of video compression schemes were described, and graphically demonstrated by Mark waving his arms wildly to show the picture update rate! Quite usable video was obtained at data rates well under 500kb/s, with all compression and decompression being done in software on Sun workstations. Some 15 other sites around the world received the audio and whiteboard parts of the lecture, and indeed there were a number of questions and comments from the network – particularly Van Jacobsen at Lawrence Berkeley Laboratory whose team are responsible for much of the technology demonstrated.

(Gordon Joly)

The view from UCL

This was the first multicast event from two sites. Brunel University hosted one end with Ian Wakeman, while Mark Handley was at UCL. There were fifteen people or so at UCL, who took part with questions to both speakers.

Mark Handley started with a general introduction to the Internet and the ideas behind "multicast". He discussed the Distance Vector Multicast Routing Protocol (DVMRP), multicast implementation and "pruning". The idea here is that receiving sites tell a sender site that they are not interested in receiving the multicast packets. There were 540 multicast routers in May 1994, of which 507 were in the US domain.

Ian Wakeman assumed that the ideas behind a multicast backbone were understood by the audience, and went on to describe video conferencing in general. He said that it was used on a day-to-day basis at UCL. It is also being used for distance learning (eg LIVE-NET). He reported a recent event involving UCL, the University of Gothenberg and a San Francisco site. At UCL it is also used in the "shared office mode" and for local surveillance to keep an eye on equipment.

He then described the H.261 bitstream standard, which was developed for ISDN and video telephony. There are both hardware and software codecs ie compression/decompression devices.

At this point, Mark Handley once again took over. He descibed the hardware devices. He looked at the various cards, including the Parcvideo card (vigrafix), and the Video capture card (videopix) from Sun Microsystems.

The Sunvideo card was the latest offering. This delivers CellB amd motion JPEG all built onto the card.

He then discussed the video processing card trade offs:

It was noted that colour dithering is a very expensive thing, and that video is not the whole story – the quality of audio is also very important.

Ian then looked at the Network implications:

Lost packets create aberration in picture, heal by processing previous and current image. If no knowledge of lost block, we need the full picture. However, this leads to problems if later blocks depend on lost blocks. This can be solved by requesting intra-coded blocks.
Constant packet generation may cause congestion and therefore packet loss can reduce the sending rate at the source, if the congestion is noted. Scalable polling of receivers to discover those which are suffering from congestion was possible. Multicast reception reports coupled with the observation that time constants must be the order of ten times the round trip time.
Hierarchical Encoding
Hierarchical video encoding implies that high bandwith gets all levels and low bandwidth only gets lower quality. It may be possible to have two different streams at different levels.
Mark talked about the Problems with DVMRP, namely that is based on scope (ttl or "time to live" – for multicast packets), and that it is a "dense mode" algorithm, and hence did not scale well. A new idea was PIM: Protocol Independent Multicast. It has two types, Dense Mode PIM and Sparse Mode PIM. These will be deployed in real routers, rather than in the present system consisting of software on a workstation. This is needed to prevent MBONE "meltdown". There is a three phase solution:
RSVP – Resource reSerVation Protocol
Receiver-based resource reservation protocol. Routes can easily change to (locally) route around a failure. It installs in a soft state and is designed with IP multicast pilot in mind. It classifies IP rather than replaces it and is an efficient classifier design. It is a current research topic and scales well to large multicasts (receiver-based).
CBQ – Class Based Queuing
It categorises traffic into classes and allocates bandwidth to each class. Lower priority classes can steal unused bandwidth from higher priority classes. It can use RSVP to install new classes for reservation.
Charging for reservation
[No information under this heading – Ed.]
There followed a demonstration of vat (Visual Audio Tool) with two video streams. Ian allowed Mark to control wb (whiteboard). Although wb was used, there were some problems with slides throughout the talk. Ian described the tool set (vat, wb and sd from Lawrence Berkeley Laboratory).

Questions were taken from Brunel first. The first questioner suggested that synchronous transmisson needs do not perform very well on IP (asynchronous) networks. Van Jacobson answered via the control wb! Ian said that the telephone lines are in fact slower, with IP taking up the full bandwidth.


(Lindsay F Marshall)

Or perhaps lack of news would be a better title. Thank you to the two people who responded to my request for ideas about possible System Administration qualifications. Come on people, we can do better than this!

And just to show you how much better we could do, Paul Anderson has sent the following report on LISA – the Large Installations Systems Administration Conference. Paul for his sins (hang on, you sin and you get a trip to Monterey?) is on the program committee for the next LISA and is looking for contributions. The call for papers should appear elsewhere in this issue so, those of you who can afford the travel, get writing as he wants to see more submissions from the UK. If I thought someone would pay for the trip I would even write something myself!

I had thought of putting up a SAGE/UK Web page, but what would I put on it? Links to other useful system administration lists? Pointers to useful software? RFCs? Suggestions, please, to me.


(Paul Anderson)

The LISA (Large Installations Systems Administration) Conference started off in 1987 as a small USENIX workshop for people interested in the special problems of managing large UNIX installations. By this year, LISA VIII had grown to a full five days of technical sessions and tutorials, with multiple tracks, and over 1200 attendees. The people present came from all sorts of sites, ranging in size from tens to thousands of workstations. A good definition of a "large site" seems to be one which has system administration problems that cannot be solved by simply scaling existing traditional solutions.

Despite a significant increase in the number and size of UNIX sites, managing large heterogeneous networks still seems to be a very difficult problem and none of the vendors seem able to provide a complete solution. Most large sites still rely on a combination of home-grown tools and procedures with an experienced systems administrator to integrate them into a complete working system. Since Systems Administration is not a traditional academic discipline, it is often very hard to locate information and learn about techniques that have been developed, so LISA provides a rare opportunity to pick up and share new ideas. The proceedings from previous conferences (available from USENIX) are an invaluable source of information for system administrators and often help to prevent "re-inventing wheels."

The theme of the eighth LISA conference was "Automation: managing the computer of the 90s." This was held in September 1994 at the Town and Country Hotel in San Diego, California. It included two days of tutorials with three days of technical sessions and invited talks. There were also the usual "Works-in-progress" (WIPS) sessions, "Birds-of-a-feather" (BOF) sessions and a vendor display. The number of parallel activities now requires some careful thought and scheduling – I tended to go to the invited talks and unofficial sessions, on the theory that the technical sessions would be well covered in the printed proceedings.

The tutorials were spread over five parallel tracks during the first two days. I did not attend any of these, but many of the presenters were well-known, experienced, systems administrators and the wide range of topics seemed suitable for several different levels. The topics included: Solaris, Security, DNS, DCE, Legal Issues, Routing, Sendmail and many others. The remaining three days consisted of one track with refereed technical presentations (with corresponding papers in the printed proceedings), and a second track of invited talks and other sessions.

The technical sessions started with two keynote speeches. The first was a business-oriented talk by Jack Stanley of the Houston Chronicle describing the migration of his newspaper systems from mainframes to "open" UNIX systems. The second was a talk about security issues by Dan Geer. Among the amusing anecdotes, he made the interesting point that good security for the Internet is vital, if it is not to be overtaken by proprietary systems, which can provide the security required by commercial applications.

Technical papers

Several technical papers (including my own) addressed the problem of installing and configuring large numbers of machines. Like our own LCFG system, GenUAdmin maintains configuration information in a central data repository and is able to do consistency checks across machines. This system generates all NIS maps and manages user information as well, but the configuration information is stored on a "static" file-basis and there seemed to be some lack of modularity and abstraction. The machines are re-configured by cron every night.

Config and OMNICONF are more basic systems for simply saving the local versions of configuration files. One paper (Riddle) described a system for automatically installing SGI systems by doing a diskless boot and automating the local disk build.

There was a lot of interest in configuring the user environment at login time. The three papers in this area included a number of interesting ideas. "Soft" won an award for best student paper. This provides a higher-level configuration language that allows users to specify which options from the standard login they want to include. One interesting feature is that this generates a shell script which is sourced by the user profile at login. The script is cached so that the login is very quick, even though the setup process may be quite complicated, and scripts can be generated automatically for different shells.

A paper from Xerox described a program for doing similar caching of the user environment at login time and a paper from BNR described their standard login procedure.

Other technical papers included:

GASH is a system which controls the editing of master NIS and DNS files. It allows authority for sections of these files to be delegated to different users and performs sanity checks on the files. (Best paper award).

The only other UK paper (from UKC) described their system for exporting single UNIX home directories to thousands of PCs using a modified pcnfsd. While not completely secure, this is a big improvement on the basic PC/NFS and it also solves the problems caused by the lack of an automounter on the PC.

Two papers addressed the popular topic of software configuration, but these seemed to be largely a reworking of previous ideas.

TenWen was a really good account of the complete rebuilding of an academic workstation network over a two-year period. This illustrated many of the general principles of good system organization.

An interesting paper from Don Libes described several mechanisms (mostly involving the expect program) for constructing programs which need to communicate passwords to other systems.

Pong is a system for pinging high level daemons at regular intervals to monitor a machine.

Other papers were:

Invited Talks

Most of the invited talks were tutorials or overviews of various topics. I found these very useful for getting an impression about subject areas that I would not normally have had time to investigate in detail.

The session on DCE was useful both for the technical overview of DCE and for the update on the latest state of OSF. DCE is a portable, secure, RPC which is available on multiple platforms (UNIX, Mac, PC, etc.) using multiple transports (TCP, Appletalk, etc.). The security has several optional levels (including full encryption of packet contents) and is based on Kerberos. X500 or DNS namespaces can be used. The COSE CDE (Common Desktop Environment) which has been adopted by many vendors (including Sun) will be DCE security compliant, so logins could be controlled by DCE security rather than standard UNIX passwords. The distributed filesystem DFS (a development of AFS) is an application based on DCE and is part of the DCE distribution.

A detailed technical talk on SNMP described the MIB structure and the information that is typically available from TCP devices. This included some interesting discussion about using SNMP for other system administration tasks.

There was also a good overview of the MBone (the "virtual network" of multicast routers that carries real-time multicast traffic over the Internet). The shared whiteboard facilities seem to offer some real practical possibilities for realtime collaborative research, as well as the possibility of audio and video. Since most routers on the Internet cannot handle multicast IP, the participating sites form multicast "islands" linked by tunneling multicast packets in standard UDP packets to a nearby MBone node. The URL htpp://www.eit.com/techinfo/mbone/mbone.html is a starting point for MBone information.

Other Sessions

As usual, the less formal sessions, such as the "WIPs" and "BOFs", provided quite a few interesting ideas, and the informal contact with so many other system administrators was very stimulating.

My general impression of the conference was that many more people are now appreciating the degree of high-level planning and organization that is necessary to efficiently run a large UNIX network. Many small tools are available for skilled systems administrators to incorporate into their own systems, but unfortunately, there still seems to be a lack of tools for genuine high-level design and management of large heterogeneous workstation networks. LISA IX Next year, LISA IX will be held in Monterey, California on 18-22 September. The theme of the conference is "New Challenges." Please see the announcement and call for papers elsewhere in this issue. Any other queries or comments about LISA in general are, of course, welcome as well.

IP Workshop and Tutorials 11-13 October 1994

(Nick Williams)

The UKUUG IP Workshop was held at London Zoo, from 11-13 October 1994, dates which happened to coincide with the 25th anniversary of the first test message being sent over the then named ARPAnet, as Peter Salus observed in his Keynote talk. The event received a good number of attendees: around 80 or so participants overall, which gave a friendly feel to the tutorials, and the consensus of agreement was that the workshop was well worth attending. Considering the topic of the workshop and the ideal venue, it was surprising that the effectiveness of RFC 1149 was not tested out, but it seems that attendees were sobered by the recent mauling of a visitor by Arthur the lion; presumably the victim was attempting a follow-up to that RFC!

The workshop included two excellent tutorials: "Network Programming" from W Richard Stevens and "Network Administration" from William LeFebvre; and also a day of talks presenting the current state, and the future, of Internet technologies.

Network Programming, W Richard Stevens

Stevens gave a very thorough tutorial, starting from how the TCP protocol works, up to the nitty-gritty of implementing some simple client/server programs. His approach of starting at what might be considered the lowest layer to some, may not have been an obvious way to go about it, but it provided an excellent grounding for the mechanisms of network programming described later in his tutorial and the method worked extremely well. Stevens is well versed in the world of TCP and network programming: the first volume of his two-volume set describing TCP has just recently been published, and his excellent knowledge of the area allowed him to present an informed tutorial which I would recommend highly.

Although this tutorial was aimed at network programmers, this would have been equally of interest to the system administrator who wants to know what's going on in the network: `just why are there so many TIME_WAIT sockets hanging around on my system?', or even to the network programmer who has been "at it" for a while, but who would like some of the magic relating to networks demystified.

The Network Administrator's Tutorial, William LeFebvre

This tutorial came last in the workshop, following all of the talks. William LeFebvre's tutorial covered the basics of administering a network of machines. The discussion of subnetting and routing carried on from the brief talk given in the workshop by Andy Davis of Spider, and Bill re-emphasised Jim Reid's comments on the advantages of using gated, although he explained how, in most simple cases, it would be easier to just install the routes manually. Bill's tutorial also combined very well with Steven's tutorial on Network programming, as many of the concepts in network administration can be more easily understood with a background of what sort of protocols are going over the network, and what the applications are attempting to do with the network. Unfortunately, the field of network management is rapidly evolving into a complicated art, and not everything could be covered in a single tutorial such as this. However, Bill gave an excellent covering of almost all areas, starting from an overview of the various IP protocols, and going on up to troubleshooting problems on the network. Probably the most lacking area was NIS, Sun's networked information service which is, despite its shortfalls, one of the most commonly used distributed systems in use today.

The Workshop

The keynote speech to the workshop was presented by Peter Salus of UUNET. Peter is currently authoring a book on the history of the Internet, using some seldom- seen material from the early days of the ARPAnet design and in this talk he presented some of the plans and memos that he's been finding lurking in dusty closets. An example he gave was a copy of the first map of the ARPAnet, showing all two nodes and the complex network required to link such a complex configuration. Those were the heady days of 1969, three years after the ARPAnet was first proposed, and 25 years before anyone would think of such silly phrases as "Information Superhighway".

Andy Davis from Spider then gave a short tutorial on the mechanism of subnetting a network: a system to provide easier network management and this was followed by Sam Wilson describing the wonders of DNS.

Keith Mitchell of PIPEX spoke about their experience of running IP over ISDN, a service which PIPEX launched last year. Although ISDN gives two to three times the bandwidth of a modem, Keith pointed out that there were still quite a few disadvantages to using it, mainly in the field of hardware/software compatability and that unless open standards are actually worked out, the success of any service provider in this field is limited. Also, with the current prices it actually works out cheaper to get a leased line unless you're using the line less than a third of the time.

From Ian Batten of Fulcrum Communications, we received an entertaining talk about the advantages of using NTP, a time synchronisation protocol to keep all of your machines to the same time. NTP provides a system for obtaining the correct time from a variety of sources, although it will work much better if you have a source yourself rather than relying on clocks distributed around the Internet. Some of the advantages of synchronised time are immediately obvious: for example make can determine correctly if a file has been modified, even if NFS is doing it's best to confuse the issue. More subtle benefits also arise. Ian commented on how suprised they were when their logfiles began to "make sense", as the distributed messages began to be correctly time ordered! The NTP software provides an excellent service and I would heartily recommend it. As a brief aside, if you're attempting to provide some sort of network security, it becomes essential to use some sort of time synchronisation system, to ensure that message spoofing cannot be performed: systems such as Kerberos refuse to work if the time difference between machines is too large.

Richard Almeida gave us a presentation on his experiences of installing and maintaining firewalls in organisations, a technique gaining in popularity, which allows access to the Internet in a controlled fashion. The idea of a firewall is that all accesses to and from the Internet pass through a single machine, which is capable of monitoring traffic and preventing unauthorised access. One of the most useful points of his talk was that the most successful way to go about installing a firewall is not to restrict services as it becomes necessary, but to restrict all services and then to enable only those which are actually used.

Jon Cowcroft described his vision of the future of the Internet, using IP Multicasting to provide a "SkyNet"-like system (not from Terminator, but from a Philip K Dick novel!). To achieve this goal, two things are necessary: an information dissemination mechanism (provided by Multicasting) and real-time reliable delivery. Jon's group at UCL has been experimenting with applications in this area for a number of years now and his talk described some of the differences in thinking between traditional network programming and programming for the "future Internet". He energetically explained how there is easily enough network bandwidth for the uses people wish to put it to at the moment and that the problems in building network applications was not increasing the bandwidth but, instead, addressing the increases in latency as the number of users increase.

Jim Reid, that man from up North, gave a presentation on gated and explained how it was the software which would solve all of the world's problems. Gated is a daemon which manages the routing tables for a machine, understanding all of the various network protocols and everything else that might possibly affect how to route messages. This is in contrast to the more traditional routed utility which tends to abuse the network, provide zero security and little actual control over what happens. The only drawback to gated is that the documentation is sorely lacking for such a complex tool, and apparently, is often wrong. However it is most definitely the routing daemon to go for.

The final talk, before the panel session, was given by Nigel Titley from BT, who presented a description of the new architecture being used to provide BT's new IP service. This service will not be available to the general public for at least another six months, but business users will be able to commence trial use at the beginning of November, making a welcome addition to the rapidly expanding market of IP service providers. The architecture being used by BT is based around an SMDS ring with "points of presence" (PoP) where required. Unfortunately, due to financial constraints, the PoPs used standard Ethernet technology, which heavily constrains the number of users which can use a single PoP. Apparently multiple PoPs will be added as demand requires.

The panel session for the workshop was thrown together at a moment's notice to discuss IPv6: the Next Generation. Unfortunately, the number of panel "experts" was severely limited (to one) and so the session ended up with Jon Cowcroft forced to taking the floor again to give a second talk. By dint of sheer good humour, Jon pulled it off, and gave an excellent summary of the IPv6 situation. Basically, the Internet has grown so much (more than the initial 200 or so hosts planned for in 1971), that we are fast running out of address space to name all of these machines (or at least, to name them in a hierarchical fashion). Because of this, a proposal was put forward to increase the size of an address in IP. The only problem was what the new size would be. Furthermore, since the basic addressing was to be changed, why not go further? The end result was the proposed IPv6, which provides for all sorts of things lacking in current IP (version 4), such as real-time priority routing, auto-configuration of host addresses and mobile computing. Most importantly, a new size of address was proposed: 16 bytes. Assuming this proposal is agreed upon, almost all other existing network protocols could be addressed from within IPv6 (apart from some little-heard of protocols from some minor organisations). The deadline to switch over to the new system is not that far away, 2005, when it is predicted the current IP will no longer be able to cope.

News from Owles Hall

(Jane Morrison)

You should find enclosed in this envelope (members only) your copy of the IP Event Proceedings, which we hope you will find of interest. Some time ago the UKUUG Council decided to circulate free of charge to all members copies of all event Proceedings as an additional service.

The recent event at the Zoo was very successful and well attended. I think everyone who attended enjoyed the venue and found both the Tutorials and the Workshop most informative.

Your Council is now looking ahead to the New Year and making plans for the next UKUUG event – World Wide Web – possibly in February or March.

Looking ahead to next year brings me to the 1995 subscription fees – you should receive your membership subscription invoice around the middle of December – this will be for the membership year 1 January 1995 – 31 December 1995 – and this is payable in advance. Once again the fees have been held at the current rates:

165 + 28.88 VAT = 193.88 GBP
65 + 11.38 VAT = 76.38 GBP
25 + 4.38 VAT = 29.38 GBP
Please pass the invoice for payment as soon as possible and if you have any queries regarding payment, or if we have the contact details incorrect, please contact us so that we can amend our records.

Although at the time of writing, it is only mid November (and extremely mild and sunny) all of us here at the Secretariat (Helen, Jenny, Jill, Bill and myself) wish you a Very Merry Christmas and a Peaceful New Year.


The Mosaic Handbook for the X Window System

by Dale Dougherty, Richard Koman and Paula Ferguson O'Reilly & Associates, Inc. 262pp + CDROM
ISBN 1-56592-095-3

(Lindsay F Marshall)

Books in the O'Reilly Nutshell series are invariable well written, well produced and timely. The publishers nearly always seem to just catch the big wave at the right moment and get a good ride home. However this book, about using NCSA Mosaic to explore the World Wide Web, may not quite give them the same run.

This does not mean that the book is some kind of total disaster. It is many times better than some of the rubbish produced by other publishers both in its production values and its content. However, there are some little points that niggle away when you look at it. The most obvious of these is that the book is liberally illustrated with screen dumps, most of which are extremely difficult to read. I thought at first this was a problem with using "Mosaic Grey" as a background on the printed page, but then I noticed that some images are perfectly crisp and clear. The problem is undoubtedly to do with the dithering involved in reducing large screen dumps to fit in the book (the colophon tells me that they used Photomatic in Adobe Photoshop). Whatever the reason the fuzziness spoils the look of the book.

Also annoying is a definite feeling of padding throughout the book. There just does not seem to be enough material to make it substantial. A tour round some of the well known Web pages is not really very interesting on paper, and, while I think of it, doesn't Mosaic have the reputation of being wonderfully easy to use, completely user-friendly, totally intuitive etc. etc.? Why then do we need a whole book on it? Shouldn't the URL of some jumping off point be enough? OK, maybe some of the Web pages are not so easy to use and need explanation, but doesn't that mean they are badly designed? So why tell us about them?

The real fun is the serendipity of following some fortuitous set of links to see where they lead and the book doesn't convey this at all. It's a bit like the Joy of Sex – even down to the fuzzy black and white illustrations!

The biggest bugbear of all is the fundamental confusion between Mosaic and the Web. Too many people already think that WWW <=> Mosaic and this book only serves to make matters worse, even though I am sure this is not the authors' intention. The only other browser mentioned by name is Lynx and that only briefly. There is also a slightly barbed reference to what was Mosaic Communications Inc before NCSA started waving lawyers about. Mosaic is a pretty good first cut at a Web browser but it is only a first step in one possible direction.

What of the technical content as opposed to the tourist guide aspects? The section on using X Resources to configure Mosaic is useful for someone who knows what they are doing, but is probably a little confusing for a beginner. The HTML material is solid but uninspiring, though to be fair markup languages are not the most interesting topic to write about. Multimedia is covered briefly, but in sufficient detail to allow someone to tweek their local system if they really wanted to. The biggest problem is that much of the material is out of date already. Mosaic 2.5 is coming soon and has differences in its interface. Extensions to HTML, like the table feature that 2.5 supports, are also showing up and the browser war that will decide the official feature set hasn't even started yet! And the hardware and software for multimedia support are changing fast too.

O'Reilly won't lose on this book – the hype surrounding Mosaic and the Web will see to that, but I suspect those who buy it will not return to it very often after their first reading. As I said before it is not a bad book, it just doesn't quite reach the standard of others in the series and I hope that the publishers won't have the gall to bring out editions for the PC and the Mac. [They already have – see news@UK, vol 3, no 5, pp 6-7. Ed.] It would all make a nice hypertext page if someone could be bothered….

Connect your Sun to the Internet for 1250 GBP

(Mick Farmer)

As soon as I heard about this product, I contacted the suppliers to ask for a review copy. Unfortunately, the final release wasn't available by the time we went to press, so a complete review will appear in our next issue of news@UK. However, I am including here some details provided by Netcraft.

Connecting to the Internet should be fun. But very often it takes weeks of effort to get all the services working properly, compiling and configuring strange software, learning by trial and error, slowly and incrementally bootstrapping your site onto the Internet.

Netcraft takes the grief and uncertainty out of getting connected. We supply a complete suite of pre-built binaries and configurations for either SunOs 4.1.3 or Solaris 2.3, together with a full source tree, a high speed US Robotics v.32bis modem, a year's subscription to Demon Internet's dial-up Internet service, a day's consultancy to get you up and running, and a month's support via Internet mail.

Clients for all the popular Internet application protocols such as gopher and World-Wide Web are provided, together with a base configuration that includes a dial-on demand PPP implementation, the latest version of sendmail, tcp wrappers protocol filter and a comprehensive news delivery system specifically configured for a dial-up connection.


link level
sendmail 8.6.9
mail user agents
elm, pine
inn+config, slurp
nn, trn, tin
dns tools
dig, traceroute
other clients
gopher, archie
mosaic 2.4
C compiler
gcc 2.5.8
build tools
patch, perl
full source tree
If you are more than two hours travelling time from Bath we reserve the right to add travelling/ accommodation costs. Netcraft will not be liable for lost time during the day's consultancy. Please ensure that you have the equipment and a suitable work environment ready for us. All prices exclude VAT.

For other operating system platforms, or for any aspect of Internet consultancy & training, please call to discuss your requirements.

X User Tools

by Linda Mui Valerie Quercia 1st Edition November 1994 856 pages, + CD-ROM
ISBN 1-56592-019-8

O'Reilly Associates has just released this CD/book set filled with hundreds of tips, tricks, scripts, and programs for X Window System users. This collection of software and hard-won wisdom is an easy-to-use reference for getting more power and enjoyment from the X Window System. The emphasis in X USER TOOLS is on useful programs, culled from the network and contributed by X programmers all over the world. The CD-ROM contains source files for all and binary files for some of the programs (for a number of platforms, including Sun 4, Solaris, HP 700, Alpha/OSF, and RS6000/AIX), as well as the software for both emacs and tcl/tk.

The programs range from fun (games, screensavers, and a variety of online clocks) to business tools (calendar, memo, and mailer programs) to graphics (programs for drawing, displaying, and converting images). The book features a collection of tips and techniques for configuring individual and system-wide environments, as well as a glossary of common X and UNIX terms.

The book uses the extensively cross-referenced style pioneered by O'Reilly's bestselling UNIX POWER TOOLS. X USER TOOLS is designed for browsing each article stands on its own, yet most contain pointers to related articles. Readers can read the book straight through from cover to cover, but most will skip around, scanning what's of interest and following cross-references to more detailed information.

Around Europe

ICT Round Table No. 8

(Mick Farmer)

The second meeting of this Round Table took place in Brussels on 27 October in one of the EC's high-tech buildings. Security was heavy and the guards took our passports in exchange for our visitors' passes (I wouldn't forget to give it back). The block was built in the shape of an arc and the front sloped backwards. This meant that our meeting room was actually curved, like a subway tunnel. Unfortunately, nobody had thought of providing similarly curved furniture. We sat around a very long rectangular table. Those sitting on the ends of the "inside" edge of the table couldn't see one another, while those in the middle of the inside edge of the table were squeezed uncomfortably against the wall. To make matters worse, the inside edge of the curved room contained the windows, which sloped inwards towards the ceiling making it difficult for the taller people to stand up behind their chairs! Finally, it was raining outside, causing condensation on the inside of the windows. This ran neatly down onto our coats and briefcases left on the window sill. Enough of such silliness. Now onto the meeting itself.

The morning was devoted to presentations from various parties. Walter de Becker started by stating that this particular Round Table (No. 8) was exceptional in that the participants had kept in touch since the first meeting (the EC still don't really understand the power of electronic mail) and that there were more people attending this second meeting than had been at the first meeting (unheard of with the other Round Tables). The increase was mainly in the number of "Service Providers" attending.

Alan McClusky (Swift Network) gave a thoughtful presentation on how he thought the "Information Society" would evolve if we started from the current Internet setup, rather than from where the Entertainment Industry would like to start; the obvious distinction being in full-duplex, rather than one-way, working. He provided an interesting analogy that within our current stage of Internet development there are no roads, only road signs.

Glenn Kowack (EUnet) gave his usual entertaining and controversial history of the Internet within Europe. Amazingly, he is doubling the capacity of his connection to the USA about every six months and that 75% of all EUnet traffic is from the USA to Europe (ftp and WWW strike again :-). The fact that 24+ countries inside and outside Europe could collaborate on such a scale without rancour certainly impressed Walter de Becker judging by his questions afterwards.

The afternoon session was concerned with the future work of this Round Table. Although it was generally agreed that the old working name "Access to Community Information for Users" needed to be widened in some way, half the meeting started a sudden brain-storming session for a new name! However, some of the more sensible amongst us agreed that this could be discussed off-line by e-mail later. Walter de Becker identified four categories of IT users that we should attempt to reach:

After considerable discussion, it was decided that the WWW was probably one area that this Round Table could address, perhaps by providing multi-organisational and multi-lingual entries. It was felt that entry points [to EC information] should be free of charge and provided by the EC. Walter de Becker summed up the EC view:
"There are tangible benefits [in using the WWW] in the short to medium term. The EC gains access to the views of an increased number of users. The WWW philosophy and infrastructure is well understood."

After he had said this, a number of other EC types suddenly voiced their agreement. In fact, we learnt that DG XIII already had a WWW page, its URL is http://www.echo.nl!

We finally agreed that this Round Table should embark on a pilot scheme to reach those people with either e-mail or Internet access. It was also agreed to investigate a partnership with a national newspaper in order to reach a wider audience. Group discussions would continue using various e-mail lists.

So, how does this help our members of the UKUUG? First, if you have any comments about the work that this Round Table is doing, please let me know. Second, if there's any IT-related information that you'd like to receive from the EC, please let me know and I'll try to acertain how to get it. Third, if there's information that you'd like to make known to the EC, especially if you're an SME (Small to Medium sized Enterprise), please let me know and I'll investigate the appropriate avenues.

EurOpen Governing Board Bucharest, November 1994

(Andrew Macpherson)

There had been rumours that this Governing Board might be interesting. EuroCheap, which the UK had been a driving force for as a way to retrench after the virtual destruction of EurOpen at Utrecht, was not taking us anywhere; yet UKUUG have a strong position of "we're paying enough" to protect. This in effect translates into "only spend on new services" as our stance for the Governing Board.

Getting to Bucharest is fun, with a lower case `f'. Direct flights via Romanian airways, or change at Vienna. Taking convenience over peace-of-mind I met Helen at Stanstead and we embarked on a 30 year old BAC-111.

Arrived at Bucharest: by the time we had the exit visas, the baggage was fairly quickly delivered. In the arrivals hall there were hundreds of Taxi-touts. Fortunately GURU had things very well organised, and were waiting for us with a bus to take us to the Hotel. Bucharest itself seemed pleasant in this quarter, with tree lined avenues, well spaced buildings and a general air of a garden suburb.

In the evening the Governing Board attendees were invited to the ROSE'94 Conference dinner in the Alexandra Palace. An interesting evening, considerably enlivened by Richard Stallman's opinions.


Took the opportunity to attend the morning session of Rose '94. Linux is making a big impact in the ex-communist countries in some ways the focus of unix-as-we-knew-it, the mutual help of the early years, has moved East.

Also an interesting talk from our chairman on distance-learning. I found the implications on what a student might be expected to have available quite startling.

Romanian Internet connectivity is over dead-slow links (a single 9600 baud line). I had a very interesting chat about various X.25 X.121 features with those running the network how these things come back to haunt one!

Governing Board

The Big Issue. Should we allow a Serbian observer in the Governing Board as a representative of a Serbian UUG, while the Serbian government does not recognise the country of one of our members (Croatia)? We voted to permit the representative of a user group to attend, provided he was not there on behalf of Serbia.

On the horizon expect two EurOpen Workshops in 1995. The first will be a security workshop to be held in London in the first half of the year. This will repeat the successful Danish workshops from the past three years. The speaker will be Rik Farrow.

The other event will be a World Wide Web conference, sometime in September, venue still to be arranged.

Our Treasurer reported that the potential conflict of interest of EurOpen vs EuNet made his position difficult, so EurOpen finds itself without a treasurer at present (just like the UKUUG).


Someone had the bright idea that we might get the job finished in good time were we to start at 08:00. Romania is two time zones East of Britain. Anyone fancy starting a meeting at 06:00?

Governing Board

The issue for today centred round the practicality of service provision at a EurOpen level, and on the successful initiative at the ICT round table (sponsored by three member groups including UKUUG). Expect some announcements in this area in due course. [See article by Mick Farmer. Ed.] Your reporter also seems to have volunteered to help the executive over the next few months. To everyone's surprise (except the chairman) the meeting finished in good time.

Tourist Bit

Presents to take back… Romania seems to be a good place to buy Vodka and Caviar. Especially if you have dollars.

It's amazing to go on a tour of the Royal Palace. This is only available to organised parties, and one must show one's passport to get in! Quite what they made of a Trade Delegation comprising two UK, one Eire, one Netherlands, one Croatian and one Russian … To get in the bus stopped alongside a 10ft stone wall with an unmarked grey door. This opened into the "customs post". From there to the palace itself we passed two video cameras and two armed guards. The palace was worth seeing.

The general tour of Bucharest took us to see the enormous building from outside, approached by an avenue to rival the Champs-Elysées. Not having seen the district as it was before the bulldozers, I was impressed, the only ugly features being the many satellite dishes on the balconies.

Return Journey

I may have mentioned this before. I don't recommend Romanian Airways. The return flight was a mere hour late in departure. A slightly more modern craft this time (Airbus 110). The real horror was Bucharest airport. Not a fun place to wait.

I was very glad to get a lift back from Heathrow. Thanks!

Across the Pond

USENIX Association 1995 Technical Conference 16-20 January 1995 New Orleans, Louisiana

The 1995 Technical Conference begins on 16-17 January with a tutorial program offering twenty in-depth courses essential to professional development. Taught by the experts in these topics, the tutorials are: Essential UNIX Programming; Advanced UNIX Security; Introduction to UNIX Kernel Internals; IP Network Administration; DCE Remote Procedure Call System (RPC); WIN32: Porting X/Motif Applications to Windows NT; Advanced Tools of Networking; Sendmail – Version 8; Writing New Motif Widgets; Perl Programming Essentials; Achieving Security in an Internet Environment; Publishing on the World Wide Web; Concurrent Object-Oriented Programming; The Kerberos Approach to Network Security; Introduction to Tcl and Tk; Multiprocessing; Topics in System Administration; UNIX Network Programming; Microsoft's OLE and the Common Object Model (COM); The Law and the Internet.

The technical sessions program for 18-20 January begins with "Ubiquitous Computing", the Keynote Address by Mark Weiser of the Xerox Palo Alto Research Center. Innovations from the Palo Alto research community are an important feature of this conference. Never-before-presented research by academic and industry researchers and developers is the fruit of the refereed presentations. These presentations, refereed by the program committee, examine essential developments in the areas of BSD, Mass Storage, Security, Objects Technology, Libraries, File Systems, and Architecture. In addition, leading experts have been invited to share their in-depth analysis of current issues in the areas of Very High Level Languages, Systems Administration, Cryptography, Operating System Measurement, Inter-Language Unification, High-Speed Networking, and Operating Systems Design and Implementation.

The Internet is a topic of great interest with talks on Internet Information Commerce, Cash on the Internet, Economics of the Internet, and the Internet at the Turn of the Millennium.

A Vendor Display on 18-19 January rounds out the conference. Software, hardware, and networking products, vendors will be ready with in-depth answers for the serious questions asked by the sophisticated if informal USENIX conference attendees.

The New Orleans Marriott is the site of all conference activities. The Marriott is offering special discounts to conference attendees.

For more information, please contact: Judy DesHarnais

Announcement Call For Participation 9th USENIX Systems Administration Conference (Lisa IX) 18-22 September 1995, Monterey, California

(Paul Anderson)

I'm on the programme committee for next year's USENIX LISA (Large Installation System Administration) conference in Monterey in September. I'm sure lots of people in the UK are doing interesting system administration work and I'd like to try and encourage some more paper submissions from the UK. The conference itself and the papers in the past proceedings are also a very useful resource for systems administrators.

I'm aware that expenses to conferences in the US aren't easy to come by, but at least the conference registration is free to paper authors.

Co-sponsored by USENIX, the UNIX and Advanced Computing Systems Professional and Technical Association, and SAGE, the System Administrators' Guild.

Important Dates

Refereed paper submissions:

Extended abstracts due: 1 May 1995
Notification to authors: 5 June 1995
Final papers due: 1 August 1995

Registration materials available: July 1995

The USENIX Systems Administration (LISA) Conference is widely recognised as the leading technical conference for system administrators. Historically, LISA stood for "Large Installation Systems Administration," back in the days when having a large installation meant having over 100 users, over 100 systems, or over one gigabyte of disk storage. Today, the scope of the LISA conference includes topics of interest to system administrators from sites of all sizes and kinds. What the conference attendees have in common is an interest in solving problems that cannot be dealt with simply by scaling up well-understood solutions appropriate to a single machine or a small number of workstations on a LAN.

The theme for this year's conference is "New Challenges," which includes such emerging issues as integration of non-UNIX and proprietary systems and networking technologies, distributed information services, network voice and video teleconferencing, and managing very complex networks. We are particularly interested in technical papers that reflect hands-on experience, describe fully implemented and freely distributable solutions, and advance the state of the art of system administration as an engineering discipline.

Tutorial Program
18-19 September 1995

The two-day tutorial program at the conference offers up to five tracks of full and half-day tutorials. Tutorials offer expert instruction in areas of interest to system administrators of all levels, from novice through senior. Topics are expected to include networking, advanced system administration tools, Solaris and BSD administration, Perl programming, firewalls, NIS, DNS, Sendmail, and more.

To provide the best possible tutorial offerings, USENIX continually solicits proposals for new tutorials. If you are interested in presenting a tutorial at this or other USENIX conferences, please contact the tutorial coordinator, Daniel V. Klein.

Technical Sessions
20-22 September 1995

The three days of technical sessions consist of two parallel tracks. The first track is dedicated to presentations of refereed technical papers. The second track is intended to accommodate invited talks, panels and Works-in-Progress (WIP) sessions.

Conference Topics

Papers addressing the following topics are particularly timely; papers addressing other technical areas of general interest are equally welcome:

Refereed Paper Submissions

An extended abstract is required for the paper selection process. Full papers are not acceptable at this stage; if you send a full paper, you must also include an extended abstract. "Extended" means 2-5 pages.

Include references to establish that you are familiar with related work, and, where possible, provide detailed performance data to establish that you have a working implementation or measurement tool.

Submissions will be judged on the quality of the written submission, and whether or not the work advances the state of the art of system administration. For more detailed author instructions and a sample extended abstract, send email to [email protected] or call USENIX at +1 510 528 8649.

Note that the USENIX organization, like most conferences and journals, requires that papers not be submitted simultaneously to more than one conference or publication and that submitted papers not be previously or subsequently published elsewhere. Papers accompanied by "non-disclosure agreement" forms are not acceptable and will be returned unread. All submissions are held in the highest confidence prior to publication in the conference proceedings, both as a matter of policy and as protected by the U.S. Copyright Act of 1976.

Authors of an accepted paper must provide a final paper for publication in the conference proceedings. At least one author of each accepted paper presents the paper at the conference. Final papers are limited to 20 pages, including diagrams, figures and appendixes, and must be in troff, ASCII, or LaTeX format. We will supply you with instructions. Papers should include a brief description of the site, where appropriate.

Conference proceedings, containing all refereed papers and materials from the invited talks, will be distributed to attendees and will also be available from the USENIX following the conference.

Where To Send Submissions

Please submit extended abstracts for the refereed paper track by two of the following methods:

E-mail: [email protected]
Fax: +1 510 548 5738
Mail: LISA 9 Conference, USENIX Association, 2560 Ninth Street, Suite 215, Berkeley, CA 94710, USA

To discuss potential submissions, and for inquiries regarding the content of the conference program, contact the program co-chairs at [email protected]

Invited Talk Track

If you have a topic of general interest to system administrators, but that is not suited for a traditional technical paper submission, please submit a proposal for a second track presentation to the invited talk (IT) coordinators: Laura de Leon and Peg Schafer.

From the Net

Munden case update

(Ross Anderson)

We managed to get an adjournment and a disclosure order against the Halifax at a directions hearing this morning. The trial will not now go ahead until at least February 1995.

The disclosure order gives me complete access to the Halifax's `computer systems, records and operational procedures'. On the basis of past experience, I would say that it is 80% likely that they will refuse to comply with this order and that Munden's appeal will therefore succeed.

However, they have lost considerable face from the incident, and it is possible that they will be stupid enough to tough it out. In order to cover this possibility, I need to recruit assistants with IBM mainframe skills (especially assembler and ACF2). I have not worked in this environment since 1989, so we need some strengthening here.

It might also be useful if we had access to a modern mainframe environment, in which – configuration difficulties always permitting – we could test code fragments if need be. We only have a rather ancient 3084 (which is due for the scrapyard next August), so we might not even be able to read a new format MVS/ESA PDS.

As for payment, it is possible to charge one's normal consultancy rate once the work starts. However, as Munden is legally aided, payment will be subject to `taxation' – this does not mean Mr Clarke, but is a legal term which means that your payment is subject to review by a special court. You might get about a half of your invoice paid, and about a year late. However, that is just one of the handicaps under which defence experts have to work in this country.

If you would like to get involved with this task, then please ship me a CV to put before Munden's solicitor. Please also let me know if you have access to a suitable mainframe, and whether you would be able to assist with visits to assorted sites in Yorkshire, where the bank's own systems are kept.


Dear Editor

Yes, I do read news@UK. The October entry on the Retrocomputing museum brought back memories of programming in Algol 60 on an English Electric KDF9 with the COTAN front end. After moving from a Friden Flexowriter paper tape system to a disk based system which had a card image model, disk space was so scarce that the "non-layout" characteristics of Algol 60 were used to flow characters to fill every character space on a card. This resulted in what only could be described as "rectangular programming". I don't wish ever to go back to those days!

A language not mentioned which brings back slightly fonder memories was STAGE2 and its bootstrap companion FLUB (first language under bootstrap). Perhaps there are still users of this language since it was by definition highly portable. It would certainly be easy to put it on to modern computers.

I found the book reviews interesting and will probably buy a couple on the strength of the reviews.

Ray Foulkes

Dear Editor

Linux SIG

I have just read the piece in news@UK, and think a Linux SIG is an excellent idea. In fact, I think this should be just a part of making the UKUUG more relevant; I have posted the following to the UKUUG news group. [uk.ukuug. Ed.]

I'm a bit worried that the UKUUG is fading away. I remember being involved in the late 80s, and there was real energy at the group meetings, with genuine excitement about the new stuff being pioneered on UNIX. It has now got to the stage where the new features of toy operating systems like MS-DOS and Windows (often cut-down rehashes of things pioneered in the open systems world) are getting all the interest. And there are even young bloods who think that playing with SYSTEM.INI qualifies them as hackers (then again, maybe it does).

Now that there are so many mature UNIX implementations around, the cutting edge interest must be outside the core of the operating system. The new enthusiasts are required to carry on the march to open systems, before they are snared by the siren voices of the corporate behemoths promoting their own proprietary software. The UKUUG needs to demonstrate that the state of the art in the global computing network is still inhabited by UNIX, and promote its position at the heart of it.

I think the first steps to revitalise interest in UNIX as an exciting technology, and publicise the UKUUG, must be to latch onto the bandwagons of today. We all know that this ludicrous "information highway" actually started in the UNIX world, but to many people it is Mosaic running on an MS Windows PC. How about publicly championing UNIX as the conduit for the Internet and World Wide Web, and latching onto the potential of Linux on the countless thousands of PCs in people's homes?

Hey, maybe I can feel a plug-and-play sampler CD-ROM coming on … get a runnable UNIX and Internet disc on the front of a major PC magazine .. interesting applications running on a UKUUG machine with on-line subscription … aarrgh … stop me before I start making sense …

I know marketing is a dirty word, but I think something like this is needed soon. System 7.5 on the Macintosh now includes TCP/IP, as will Chicago, the next release of MS Windows. The big boys have finally realised that people want global connectivity; the time must be right to show which operating system does it best.

David Whitwell

[email protected]