Reminder & Update - ISMB '94 a/v broadcast on MBONE

Dan Mosedale mosedale at aeffle.Stanford.EDU
Tue Aug 9 19:33:35 EST 1994

We would really like to have as many folks take part as possible;
please forward a copy of this announcement to any friends, colleagues,
or mailing lists you think might be interested.

If you are in a university setting and don't have time to get your
machines on the MBONE, try calling up the networking division, the
computer science department, and/or the computing center.  They may
already be connected and have a machine available that you could use
for viewing.

Finally, this announcement contains a newer version of the "Quick and
Dirty Guide to Connecting to the MBONE" which has some important
revisions from the previous version posted.



		The Second International Conference on
              Intelligent Systems for Molecular Biology
                         Stanford University

  Live AUDIO, VIDEO, and WHITEBOARD broadcast over the Internet MBONE

			  August 15-17, 1994

		   8:30 AM - 5:30 PM,  PDT (UTC-7)

The ISMB conference is intended to bring together scientists who are
applying the technologies of advanced data modeling, machine learning,
artificial intelligence, robotics, parallel computing, and other
computational methods to problems in molecular biology.  

We plan to broadcast the refereed paper presentations as well as those
of the invited speakers over the MBONE.  Those folks not able to make
it to the conference in person can still watch live video and audio of
the presentations on an Internet-connected UNIX workstation.
Furthermore, Internet viewers will be able to participate in question
& answer sessions via audio (microphone required) or email.

More information about the conference itself is available on the
World-Wide Web at
If you don't have access to Mosaic or a similar WWW viewer, you can
get this file by anonymous ftp and simply read through it.

In order to view the conference, both your workstation and network
need to be configured properly.  Someone with at least some small
amount of UNIX system administration experience ought to be able to
get this going with the investment of somewhere between 20 minutes and
6 hours of work.  Typically, this will be spread out over a few days
or weeks, as you will probably need to exchange some email with your
network provider.

As such, if you are interested in watching or participating, now would
a good time to get started.  I've attached my "Quick & Dirty Guide to
Getting Connected to the MBONE" to this posting.  Give this to your
local system administrator and see if he or she is interested/willing
to do this for you.

The conference will be announced in sd (the multicast session
directory) shortly beforehand.

Thanks in advance to Digital Equipment Corp., as they will be
providing us with the necessary computer hardware to make this happen.


Author: Dan Mosedale <mosedale at genome.Stanford.EDU>
Last updated: 94/08/09
Version: 0.8

This document copyright 1994 by Dan Mosedale.  

Comments, corrections, and improvements are welcomed. will always contain
the most current version of this version of the document.

Recent changes to this document are marked with a vertical bar
character | at the beginning of the changed lines.

Dan's Quick and Dirty Guide to Getting Connected to the MBONE

* First, get and read 

    "MBone Provides Audio and Video Across the Internet"
    Michael R. Macedonia and Donald P. Brutzman
    _IEEE COMPUTER_, pp. 30-36, April 1994. 

    PostScript, text, and hypertext versions of this article are available as

   This is a good introduction to what the MBONE is and how it works.

* READ THE MBONE FAQ LIST, available at

    Note to network admins: the version of the FAQ available at AT&T has one 
    extra section that talks about the security implications of multicast 
    packets, including a few issues involved in tunnelling them through a 

    The rest of these instructions assume that you have read the FAQ.

* make sure that your network provider can supply a multicast feed

  Call your network provider/manager.  If they don't already get MBONE
  packets but are willing to try and get them for you, point them to a
  copy of the MBONE FAQ, which explains where they can find out about a feed.

* subscribe to the mbone mailing list by sending a message to 

     <mbone-request at>

  and asking to be added.  If you have any problems while trying to do
  the install, or if you can't find multicast binaries for your
  machine, chances are that someone else has had the same question.

  The first thing to try is to download the archives
  ( and browse/search through
  the file to see if it's already been mentioned on the list.

  The second thing to try if you should have problems is to browse
  through various MBONE resources available via the World Wide Web.
  http://genome-www.Stanford.EDU/~mosedale/mbone.html contains a
  meta-directory of many of them.

* configure machine(s) for multicasting 

  How much work you need to do to get multicasting going depends in
  large part on how modern your OS is.  Notably, Solaris (2.1+),
  BSD/386 (1.1+), DEC OSF/1 (2.0+), and IRIX (4.0+) come with kernel
  support in place.  
  From here on, I will assume that you will be running your multicast
  applications on the same machine that you use to run mrouted
  (assuming your LAN needs a tunnel).  If the machines you have are
  all slow, you may want to configure two machines for multicasting
  and split the duties.

  As far as kernel multicast support, fixes, and application binaries
  for various boxes:

    WARNING - I have not tested all of the platforms listed here and
    thus cannot personally verify that they all work.  Further, some 
    applications may not be available for every platform.
    OTHER SOURCES - in this document, I have listed the primary
    ftp sites for most programs.  However, almost all of the programs
    mentioned here can be found at other archive sites which may be
    closer to you.  Please check the closest sources first.  Archive
    sites include (UK), (Germany), (Australia), (Japan).

|   SunOS 4.1.x
|       Available at
|       If you are running SunOS 4.1.3_U1B, you will want to get both
|    	version 3.1beta along with 3.1beta.patch1.  Otherwise, get
|       ipmulti-sunos41x.tar.Z.
|       To install these patches Ignacio Martinez <martinez at>
|   	gives this tip:

        > Configure kernel and patch kernel code with the sources
        > included in the distribution (the important thing here is
        > not just copying the object files *.o and rebuilding, but
        > generate new *.c & *.h files from the patches as well)

        SPARCstation 5 note: in SunOS 4.1.3_U1B, there are kernel bugs
        related to the audio device.  Even after applying jumbo patch 
        101508-06, only some of the bugs are fixed.  Rumor has it that
        Sun is aware of said bugs and that a complete patch is under
        construction.  In the meantime, you will probably want to use
        a different machine.

|   Ultrix (DECstation MIPS)
|       Source patches and binaries are supplied for some versions of
|       RISC Ultrix at  There
|       have been conflicting reports as to whether these actually work
|       with Ultrix 4.3.

        Packetfilter Note: if you are using the Ultrix packetfilter on
        your system (with CAP or Netman, for example), installing the
        multicast patches may break the packetfilter.  If you are
        running Ultrix 4.2A or 4.3, you can get a version of the
        packetfilter which does work with multicasting at


        v4.0.x has mrouted and kernel support included, but both are badly 
        broken; you must get and install the patches from 

        v5.x includes kernel support and mrouted.  Newer versions of 
        mrouted et al can be found at; 
        these are recommended but not required.  

        Also, Ran Atkinson <atkinson at> says:

        > There is a bug in SGI's X server software that interacts
        > badly with wb.  A workaround for the bug is to run xpsview
        > once upon login and before attempting to run wb.  SGI is
        > aware of the problem and says they will fix it in some
        > future release of IRIX.  The bug is reportedly present in
        > IRIX 5.2 and earlier.

    HP/UX 9.01 (9000/7xx)
        contains kernel patches and mrouted.

    Solaris 2

        Working kernel support is included in Solaris 2.3.  Earlier
        versions may require a patch.  The directory contains
|       this patch, along with all the application binaries.  Be sure
|       to set the appropriate flags to ifconfig (see the man page).

    DEC OSF/1 (Alpha)

        v2.0 includes kernel support for multicasting, but not
        mrouted. ftp://genome-ftp.Stanford.EDU/pub/mbone/mrouted2-osf.tar.Z
        has both sources and binaries.

        v1.3 requires kernel patches as well as mrouted; get them at
|       However, these bits are a beta-test version, and are not
|       supported by DEC.  Furthermore, the current versions of many
|       precompiled application binaries (vat, wb, etc) only support
|       OSF/1 v2.0.  In short, you are encouraged to upgrade to OSF/1
|       v2.0 if at all possible.  
|	If you absolutely can't avoid using a DEC OSF/1 v1.3 machine
|	as your mrouter, here is a tip from Heather Gray <heather at>:
|       > The problem (at the time of the V1.3 patches) was a bug in
|       > the lance driver that resulted in the ALLMULTI mode, which is set
|       > when mrouted starts, being subsequently turned off by the driver,
|       > meaning that new memberships weren't heard.
|       > The workaround to this was to ifconfig the lance to set the address
|       > and mask and bring it up, start mrouted, then ifconfig the lance 
|       > interface again just using the "allmulti" option
|       > (e.g. 'ifconfig ln0 allmulti'). 
|       > V2.0 doesn't have the driver bug so there is no need to use the
|       > additional ifconfig.

    BSD/386 1.1 (Intel 386/486/586)

        From the vat README file:

        > Not all of the IP Multicast changes made it into BSD/386 v1.1.
        > In particular, the in_pcb fixes that bring the network code
        > into conformance with the Host Requirements RFC were left out.
        > Apply the kernel source patches in
        > if you'd like your kernel to properly demultiplex
        > multicast packets.

        If you don't have kernel source code, you can find object files at
        contains most of the necessary application binaries (nv, sd,
        mrouted, vat, etc.).

        The machines listed below can be patched to support multicasting, 
        but are missing most of the necessary applications (ie sd, vat, 
        and wb).  So you will probably want to avoid trying to use these
        machines unless you are prepared to volunteer to port said programs.

    AIX 3.2.5 (RS/6000)

    NetBSD 0.9

* get sd, vat, wb, and the lbl version of ghostscript from the subdirectories
  of  Note that if you already picked up 
  the binaries for these somewhere else but don't have man pages, you can get
  them here also.

    - install them as per the instructions

|   - although wb supports Display PostScript as an alternative to
|     GhostScript, you may want to use GhostScript anyway: many slides
|     transmitted using wb are sent using the lz-compression filter to
|     save net bandwidth.  Only the very newest versions of Display
|     PostScript support this, and many vendor X servers don't yet
|     have this (SGI IRIX v4, DEC OSF/1 v2, probably others).  If you
|     are using DEC OSF/1, get GhostScript 3.0 (the first version to
|     fully support the Alpha architecture) from

    - note that for some machines (notably Digital boxes), you will
      also need to ftp and install AudioFile in order to use vat.
      Details are supplied in the README file for vat.

* get nv from

    - binaries for many machines are available as nv3.3beta-*.  For 
      those who prefer to build their own binaries, source code is
      available there as well.  Note that this is a beta-test
      version; however, it's been extremely stable in the couple of
      months that I've been using it.
    - install it as per the instructions

* with help from your net-provider, configure a tunnel using mrouted 
  (if someone on your network hasn't already done this)

    - Using the information in the man page for mrouted, the FAQ list
      (which has a specific section about this), and help from your
      network provider, configure a tunnel to put multicast packets on
      your LAN.

* test sd 

    - within a minute or so (maximum) of starting up sd, you should
      see at least six sessions, often times a lot more.  If not, wait
      a while and try again (very occasionally, the entire MBONE loses
      connectivitiy).  If this doesn't do it, check out the mailing
      list archives and www resources mentioned above.  If you don't
      find anything there, send mail to the mbone list.

* test vat

    - on some machines, you may need to start the AudioFile server at
      this point.
    - in sd, select the "MBONE audio" session.  This will start vat.
      You should quickly see the names of 20 or more people; there are
      always this many hanging around.  Select the help button for
      some info about how vat works.

    - if you have a microphone, turn it on.  Use the procedure
      described in the help window to ask if anyone on the net is
      hearing you.

- test nv
    - in sd, if there are any conferences going on which advertise
      video in nv format, click on them.  You should see pictures.

    - if there are no such conferences, click on "MBONE video."  If
      you have a mike, you can ask the "MBONE audio" session if
      someone can send you some video for testing.  Most of the time,
      nothing is being broadcast on MBONE video.

- test wb

    - if you are on an SGI machines, and are running IRIX 5.2 or
      earlier, now is the time to start xpsview and then quit.

|   - if you are using GhostScript, be sure that you have the following
|     X resource setting in place:
|	wb*UseDPS:	false
|     otherwise, wb may try to use Display PostScript, which may not have
|     the lz-compression filter.

    - select an sd session that includes whiteboard media.  Wb is
      pretty self-explanatory -- look it over.  Participate, if

- a note about transmissions
|   First, a note from Van Jacobson <van at>:
|   > At this particular point in time, the mbone offers a unique
|   > research opportunity to explore what is possible when people are
|   > able to distribute audio and video content that could not or
|   > would not be handled by traditional media -- things like
|   > interactive conferences, seminars, classes, meetings, etc.
|   > Because of the very limited bandwidth resources, the mbone
|   > community has been very careful to adopt a `greatest good for
|   > greatest number' philosophy when planning events -- i.e., things
|   > that would only interest a small community are deferred or
|   > canceled if they would compete with things of wider interest.
|   > The mbone has a total of about 500kb/s of bandwidth that has to
|   > be shared between 1200 networks and 10,000 hosts.  To put that
|   > in perspective, 500kb/s is a *total* of 6 vat conversations or 4
|   > nv video sessions.  6 conversations shared between 10,000 hosts
|   > is not very many (and because of the multicast scoping rules,
|   > just because you can't see other conversations, it does not mean
|   > you aren't interfering with them).  There are many worthy,
|   > useful things happening over the mbone and more that can't
|   > happen yet because there isn't enough bandwidth.  With so much
|   > good content competing for so little bandwidth, it is a massive
|   > disservice to the community to be filling up our precious
|   > bandwidth with junk.
|   So before you let your entire network loose to play on the MBONE,
|   please make sure that they understand this.  If you wish to
|   multicast a wide-area session, PLEASE re-read the section of the
    FAQ which relates to TTLs and thresholds, as well as the man page
    for mrouted and the man pages for the applications you will be
    using (vat, nv, etc).  If you don't know exactly what you are
    doing, it is very easy to flood large chunks of the Internet with
    massive amounts of data.  Very often, when this happens, the folks
    who made the mistake have no idea that they are causing a problem
    until they receive a torrent of angry email.

This is it -- you are now connected!

Dan Mosedale, Systems Admin            Email: mosedale at genome.Stanford.EDU
Stanford Genetics Department

More information about the Bionews mailing list