Recommendations to the University of Manitoba Task Force
on Strategic Planning
All we are saying is give X a chance.
The high cost, complexity, lack of intercompatibility and rapid obsolescence
associated with the PC model of computing is a cronic problem that can
only get worse as the computing capabilities needed in all fields evolve.
Network computing promises substantial cost savings, while at the same
time providing more sophisticated and reliable computing resources to the
user community, along with easier use and insulation from obsolescence.
This Web site and the sites to which it is linked can make a compelling
case that the Network Computing model will largely replace the PC model
over the next decade. The factors of lower cost, ease of administration,
and convenience for the user will drive this shift.
The University of Manitoba has Network Computing experties second to
none at Canadian Universities. It is therefore in a unique position to
become a pioneer in Network Computing. The NC model is already quite mature
on our Sun Unix system, although few users use it to its fullest potential
at the momemt. Making the NC model widespread on campus is therfore a matter
of expanding the current system, making it a bit more user-friendly, and
providing NCs for students and staff. Network
Computing is a area in which the U. of M. can easily stand out as a leader,
which can be a valuable selling point for the recruitment of both staff
and students.
I propose the implementation of the following steps to make it easier
for users at the University of Manitoba to migrate to the NC model:
- Phase out investment in PCs in public areas, such as student labs
and ACN open areas, in favor of the purchase of NCs. It is instructive
to note that in student labs such as the ACN open area in Agriculture,
many PCs sit idle even during peak afternoon use periods. At the same time,
the NCs and workstations in the ACN open areas are chronically full every
weekday afternoon during classes. Most open areas such as Agriculture,
don't have any NCs. Current plans to add more PCs to these areas therefore
have little justification.
The more rapidly we shift to the NC model, the greater the savings.
In fact, the best way to pay for the switch to the NC model is by cutting
back on expenditures on PCs and LANs. If this is done, it should be possible
to move to the NC model with no new capital outlays,
and perhaps some savings, even in the very first year.
- Add NC capability to existing PCs. Currently, most open area
PCs have a shareware program called microX, which allows the PC to function
as an NC. However, the configuration needs to be improved so that the user
gets the same desktop at login, regardless of whether he or she is logged
into a NC, a workstation, or a PC. microX has many limitations that aren't
strictly compliant with X11 protocols. It might be better to look into
bulk licenses for one of the commercial X-window packages. Such a package
should include low cost licenses for purchase by students and staff to
use this software on their home computers.
- Simplification of the process of running an NC from home. ACN
has already done an excellent job of providing high speed modem access
for SLIP or PPP protocols to users at home. However, if a little fine tuing
may be in order to "idiot proof", as much as possible, the process
required to set up X11 capability on one's home computer. In the past,
expansion of modem capacity has been managed by charging nominal use fees,
which should probably work well in the future.
- Expansion of server and network capacity. Probably the most
cost effective strategy is the purchase of Sun Enterprise hardware, which
greatly simplifies plugging in addtional CPUs, disks, and other devices.
All future investment in server technology should be 64-bit CPUs, in anticipation
of the switch from a 32-bit to a 64-bit operating system. There may be
some need for increasing the capacity of the campus backbone to handle
network traffic. Departments, Faculties, or user
groups, or individuals that need to have dedicated CPUs or workstations
could have those processors integrated into the main system.
- Convert the existing LAN servers into NT servers to be used for
running Windows applications via Wincentre
. Obviously, the University has a substantial investment in LAN
servers. Conversion of these LAN servers into NT servers preserves, or
perhaps enhances the value of this investment.
- Development of a comprehensive server-based software desktop. The
existing Sun/Unix system has almost all of the components of a complete
PC-style desktop already in place, including word processing, full internet
capabilities, and graphics and statistical packages. Additionally, there
are many specialized reasearch or teaching oriented facilities such as
BIRCH. The
goal should be that the typical user can do all computing tasks on a single
system, from their home directory. Applications could run anywhere,
whether on the Sun system, on some other Unix box, on a VAX or IBM mainframe,
or a Windows NT server, but they would display on the same desktop. In
many cases, it should be transparent to the user where these applications
are actually running. Most of the critical pieces are already in place.
However, there are a few things that need improvement:
- Existing WordPerfect licenses should be upgraded to 7.0.
- Where Unix versions of critical software packages do not exist, these
shoudl be made available on the CDE
desktop through Wincentre
running on an NT server.
- The current Unix system lacks good spreadsheet and user-friendly database
packages. In the short term, Windows versions of programs such as Microsoft
Excell and Corel Quattro Pro should be integrated onto the desktop via
Wincentre. As JAVA versions of these programs become available, they shoud
be installed.
- A mechanism needs to be in place to provide course-related software,
most of which exists only on the MS-Windows or DOS platforms. It should
be straightforward to run these programs on NT-servers via Wincentre.
- Web-based documentation of how to use the unified system should be
placed in an obvious place under UMInfo. There should be a straightforward
beginners learning path that introduces the user to the system, details
how to set up one's account, and where to go for more in-depth information.
- Administrative units should continue migrating to a client/server
model for distributed computing. Most administrative computing is already
largely centralized. X11 or Web-based front ends should be emphasized for
future development of these systems.
- Negotiation of discounts for bulk purchase of Network Computers
and NC software for PCs. Because network computing is only now beginning
to become widespread, there may be opportunities ot get discounts and perhaps
even gifts of hardware and or software from vendors who are interested
in publicizing the advantages of network computing.
Return to Introduction to Network Computing