[comp.unix.large] Large systems

ikluft@uts.amdahl.com (Ian Kluft) (09/12/90)

In article <barn.652971765@convex.com> barn@convex.com (Tim Barney) writes:
>I suggest that any UNIX is a large OS. Networking code seems to add quite
>a bit, and then there's all the "frills" we take for granted nowadays [...]

Yes, Unix is a large OS.  But not all Unix systems experience the problems
associated with a large system or network.  If a system and network can be
administered by 1-3 people, it's staff probably will not have a grasp of what
it means to run a large system.

However, I don't think we should try to put numbers on what a large system is.
For example, 300 users, 500 terminals, 200 GBytes DASD (as disks are called in
mainframe-land), 64 MB RAM, 32 I/O Channels, 80 workstations, 5 subnets, etc...
That's a big system.  It's also an unrealistic threshold because administrators
can experience "large system" challenges before many of those criteria are
reached.

>Now, who wants to discuss these issues concerning LARGE systems (even ignoring
>the absence of a definition for LARGE)? Some one must have wanted to, because
>these newsgroup never get created without some voting. Maybe there's a lot
>of LARGE UNIX system spectators waiting for a game to start?....... :-)

**** Here's the point ***

We don't need a definition of "large".  If a system administrator is having
problems because old methods don't scale well to a current or future system,
this is the area to discuss the problem.  More generally, I believe that we're
discussing problems with parts of Unix, administration methods, and applica-
tions that don't scale well to today's mainframes, supercomputers, and networks.

That seems to be the broadest definition that fits a discussion of "large
systems and networks", as listed in the voting for this newsgroup.
-- 
signature follows

Ian Kluft                                           ==            ==         ==
UTS Systems Software                                ==            ==         ==
Amdahl Corportation    =======  == ==== ====   =======   =======  == ======  ==
Santa Clara, CA       ========  ============  ========  ========  =========  ==
                            ==  ==   ==   ==  ==    ==        ==  ==     ==  ==
ikluft@uts.amdahl.com  =======  ==   ==   ==  ==    ==   =======  ==     ==  ==
                      ========  ==   ==   ==  ==    ==  ========  ==     ==  ==
 Amdahl mainframes    ==    ==  ==   ==   ==  ==    ==  ==    ==  ==     ==  ==
are the largest Unix  ========  ==   ==   ==  ========  ========  ==     ==  ==
systems in the world  ===== ==  ==   ==   ==  ========  ===== ==  ==     ==  ==

richard@pegasus.com (Richard Foulk) (09/14/90)

>signature follows
>
>Ian Kluft                                           ==            ==         ==
>UTS Systems Software                                ==            ==         ==
>Amdahl Corportation    =======  == ==== ====   =======   =======  == ======  ==
>Santa Clara, CA       ========  ============  ========  ========  =========  ==
>                            ==  ==   ==   ==  ==    ==        ==  ==     ==  ==
>ikluft@uts.amdahl.com  =======  ==   ==   ==  ==    ==   =======  ==     ==  ==
>                      ========  ==   ==   ==  ==    ==  ========  ==     ==  ==
> Amdahl mainframes    ==    ==  ==   ==   ==  ==    ==  ==    ==  ==     ==  ==
>are the largest Unix  ========  ==   ==   ==  ========  ========  ==     ==  ==
>systems in the world  ===== ==  ==   ==   ==  ========  ===== ==  ==     ==  ==
 ^^^^^^^^^^^^^^^^^^^

How about the largest waste of network bandwidth per article.

-- 
Richard Foulk		richard@pegasus.com