June 18, 1997

 

Report of the Subcommittee to Review Cost Recovery for Dial-In Access

 

Members:

 

Mr. Robert Cook, Mr. Kevin Harris, Mr. Eugene Siciunas, Dr. Eva Swenson, Mr. Paul Tsang,

Professor James W. Wells, Professor Robert M. Wright: Chair

 

 

1.0 Introduction:

 

As the University of Toronto moves towards better integration of electronic services into teaching, research and administration, the issue of access and delivery of these services becomes an important part of academic planning. One aspect of the delivery of these services is the provision of remote access to the campus via a dial-in service (UTORdial) now provided by the University. This service has been extremely successful and is now part of the services provided that many have come to expect, take for granted and have assumed that the University will continue to support out of base budget. As is prudent, the University has now undertaken to review this service to evaluate its use, effectiveness, costs and to examine its methods of delivery and of financial support. It is important to state from the beginning that this subcommittee's mandate was to examine, discuss and make recommendations specific to Dial-In Access only.

 

At present, the Dial-In service is provided by the University through base budget contributions which currently allow faculty and staff to use the service for "free" (individuals not charged directly) and is also supported by charges to students (after some free access i.e. a one time free token).

The idea that the University might begin to change the basic administration of the system for example, charge faculty for Dial-In Access or out-source this service (privatization and associated costs to the individual) has stirred some strong emotions and opinions. Briefly stated the issues are:

1.	The perceived need by many faculty for the ability to access the campus remotely as part of the requirements of a faculty member's "job" and the perception that the University as a whole benefits from having faculty and students use Dial-In Access for academic purposesÉ..more "Scholar for Dollar" to put it crudely.
2.	The perceived inequity between faculty receiving the service apparently for "free" and students explicitly having to pay.
3.	The perceived inequity between faculty that work on campus and those that work at home and need remote access to carry out their University work.
4.	The issue that whatever access services the University supports have a cost of implementation, maintenance and delivery. Demand for Dial-In Access is high given the initiatives the University has initiated e.g., the Information Commons, other Library Services and use by Faculty of the Internet for teaching. The expected increase (starting in 1999) of services that will be available to students over the Web such as accessing biographical portions of their academic record, or enrolling in courses. Costs of delivery and long term support are an important issue.
5.	The perceived problem of "inappropriate" use of the access time available by either faculty and students due to long log in periods "parking" or use by persons not employed by the University of Toronto e.g. family members or friends.
6.	The fact that some faculty, programs or divisions are now not currently on the University backbone and therefore must use Dial-In Access for Internet communications.
7.	The concern that "free" access leads to unlimited demand and therefore inefficient use of network services, difficulty in meeting demand, and to an unwillingness of departments or divisions to upgrade to the University's backbone, thereby avoiding costs.
8.	The perception as to whether Dial-In Access is a "fundamental" part of academic delivery at the University, or a value added service which should be treated with a different service delivery mechanism, i.e., cost recovery formula.
9.	The appropriate balance of service delivery to costs for a valued resource.
10.	The issue of whether a private Internet service company could provide better, more cost effective service to faculty and students.

 

2.0 The Existing Dial-In Service

 

The UTORdial facility currently consists of:

 

128	Modems and Ports at 14.4 Kbps
144	Modems and Ports at 28.8 Kbps (soon to be upgraded to 33.6 Kbps)
48	Modems and Ports at 33.6 Kbps
___
320	Total

 

There is another module of 48 modems and Ports on order in anticipation of continued growth in demand for Dial in Services. Phone lines from the soon-to-be-decommissioned PACX (70 lines) will be used to provide the necessary lines.

 

 

3.0 Existing Use and Charges

 

3.1 Growing Use (Tables 1 - 12 are found in Appendix 1)

 

Tables 1 and 2, titled "System Accounts Summary Report" for April 14, 1997 and April 15, 1996, respectively depict the number of users in each account class, and their activity in the past month and week, and the number of new accounts created in the past month and week. While the absolute number of accounts continues to rise, more than doubling in this period, the number active in a given month is growing marginally more slowly.

 

Tables 3 and 4, titled "Weekly Statistics" for the weeks beginning April 7,1997 and April 8, 1996, respectively show the usage by each class of user, as well as by modem pool dialed, and actual connection speed. In 1996, the 14.4 Kbps modem pool was more popular than the 28.8 Kbps pool but, in 1997, the roles are reversed. It is significant to observe from the Account Summary Tables, that while there are twice as many active students and graduate students as there are faculty and staff, the faculty and staff consume a comparable amount of connect time. Of the 32,070 connect hours consumed in the week of April 7, 15,614 were chargeable, since they were used by alumni, students and graduate students. The rest were consumed by faculty, staff and support personnel. Moreover, the number of calls made by faculty and staff, 32,332, versus the 54,449 by alumni and students, indicates that the faculty and staff call in less frequently, but stay connected much longer than do the students. This behaviour is consistent with the supposition that a "free-good" is more readily consumed than one that is charged for. While some of this greater consumption is due to good and proper necessity, there is also the potential and likely the reality of abuse of the facility by idle or frivolous use, or by other than the university member.

 

Tables 5 and 6, titled "Number of Tokens" and "Weekly Statistics", show the history of usage, since April 29, 1996. Table 5 shows the seasonal variation in the consumption of tokens by students and alumni, and hence revenue. The important fact to note is that this summer's level of use by the students is approximately double the level last summer. Table 6 shows the history of variations in total connect time and the number of calls. Again the significant observation is that usage levels are more than twice what they were last year.

 

 

3.1 Evidence of Congestion and Proposed Metric

 

In order to quantify the congestion that was being observed on the UTORdial service, Bell Canada was requested to provide a report on the number of times that a busy signal was encountered, on an hour-by-hour basis for the week of April 13 to 19, 1997. This data was then compared to internally-generated statistics in order to arrive at a meaningful metric that could be used to signal the need for additional lines and modems. The Bell report is shown as Table 7. It indicates that for most of a day, up to approximately 1000 or so connections per hour can be made without encountering a busy signal (indicated by Overflow in the table). With 320 lines, this suggests users stay connected for an average of about 20 minutes. During the early morning hours just after midnight, a lesser number can be accommodated, which is indicative of these users staying connected for longer periods of time. During the peak evening hours, of say, 21:00 on Monday, 3705 busy signals were logged, out of 4555 attempted connections. This means that 850 were successful. It is impossible however, to determine how many individuals are represented by the 3705 busy signals. It appears that many users configure their modems to re-dial immediately upon encountering a busy signal, so the true impact on users cannot be determined from these statistics.

 

An examination of statistics provided by the UTORdial system itself, when compared to the Bell data can yield a meaningful metric. Monday at 21:00 is an example of the worst congestion, while Friday at 23:00 is indicative of the onset of congestion. Comparison with the graphs on Tables 8 and 9, which depict the Saturation of the 28.8 pool on these days respectively, confirms the Bell data. Such graphs, however do not constitute a readily quantifiable metric that can be used as a threshold to trigger the addition of more lines. A different view of the statistics generated by UTORdial is shown in tables 10 and 11, again for Monday and Friday, respectively, showing the activity in hours per line. As expected, Table 10 shows that the last line in the hunting group was occupied for approximately five hours on the Monday, consistent with the data in Table 8. On Friday, however, the last line was only occupied for about an hour, and relatively few busy signals resulted. Thus a useful metric would be the onset of use of the last line, for say, more than half an hour a day. From Table 12, which tabulates, among other things, the number of hours above a threshold of 95% busy, the appropriate figure for the Friday that correlates with the onset of congestion is 0.27 hours at 95% busy. This can then serve as an automated metric, with a threshold of say, 0.2, which if reached for more than two or three consecutive days, to eliminate anomalous situations, can trigger the addition of more lines.

 

 

3.2 UTORdial Project Funding

 

Table 13 below shows the funding status of the UTORdial service, as of May 1, 1997. This shows the sources of funding, namely the Provost's contribution and revenue from token sales, as well as all the expenses incurred since the inception of the service in September, 94. Essentially, there is a funding surplus of $5,098, with a $70,000 capital loan repayment yet to be made in 97/98.

 

Table 13 (a): Sources of Funds:

Token Sales	95/96	58,163	Sub-Total
	96/97	264,658
			322,821
Base Budget for Phone lines	95/96	72,000
	96/97	72,000
			144,000
Central Funding for UTORdial	95/96	372,180
			372,180
Loan from UTC Transition Fund	95/96	140,000
			140,000
			Total Funding
			979,001

 

Table 13 (b): Expenditures:

320 Modems and Ports		484,309	Sub-Total
Port and Modem maintenance		33,080
			517,389
320 Phone line installation		24,275
Phone line charges	94/95	26,520
	95/96	73,440
	96/97	101,707
			225,942
Project Contract labour		91,572
Ongoing Support (NDIS+Ops)		69,000
			160,572
Loan repayment	96/97	70,000	Total Expenses
Total Project Expenses	To May 1, 97		973,903
			Balance
Balance of Funding			5,098

 

3.3 Projected UTORdial Budget (from May 1, 1997)

 

Table 14 below shows a projected budget for '97/98, under the assumption that the revenue from token sales stays the same as in '96/97. Clearly, this is a very tenuous assumption, since the decisions made as a result of this sub-committee's recommendations will have great bearing on the revenue and hence the budget outcome. The net result is that if the revenue stream can be maintained, the UTORdial capital and operating costs will have been covered. If no more changes or enhancements to the service were made after '98, then the ongoing annual costs would be $222,024, which after the Provost's contribution of base budget of $72,000, means that $150,024 would have to be earned from token sales. Of course, any additional modems and lines would add to these costs, as would any conversion of 14.4 Kbps modems to 33.6Kbps.

Table 14: Projected '97/98 Budget

Sources of Funds
Balance from 96/97	5,098	Sub-Total
Base Budget for Phone Lines	72,000
Token Sales: (Assume same)	264,658
Total Expected Funding		341,756
Expected Expenditures
320 Port + Modem Maintenance	35,452
320 Phone lines	115,200
Ongoing Support (NDIS+Ops)	69,000
Loan repayment '97/98	70,000
48 More Modems + Ports *	48,732
48 Line Installations: **	3,641
48 Port + Modem Maintenance: ***	1,186
Total Expected Expenses		343,211
Projected Balance at end of '97/98		(1,455)

Notes:	*	- On order
	**	- No ongoing expense for phone lines, since budget for existing PACX phone lines would be used.
	***	- Assume installation date of Aug 1, 1997, 90 day warranty

3.4 Future Service Provision and Costs

The incremental costs of adding additional modems, ports and lines, in units of up to 48, are shown below.

Item	One Time	Continuing
48 Ports and Modems (Cisco 33.6Kbps)	48,732	2,372 *
48 Phone lines and installation	3,641	17,280
Total	$52,373	$19,652
After two more 48 port modules, rack	$800
Note: * - annualized maintenance, after 90 day warranty

No additional support staff are expected to be required for expansion to foreseeable sizes. Expansion beyond approximately 1000 modems and lines may require additional resources and server facilities. Due to ever increasing technological change, it will be important to continuously evaluate the University's service against what level of service and costs can be provided by outside suppliers. It is anticipated that the service will not expand beyond a 1000 modem pool at the 33.6 Kbps speed.

 

 

4.0 Dial-in Services and Charges at Selected Institutions

 

Item	U of T	McMaster	U Waterloo	York	Queens U	Guelph U	Carleton U
#Fac/Staff@	6,198	2,302	2,438	3,207	2,190	2,093	1,709
#Students	50,000	20,000	20,000	40,000	17,000	13,000	16,000
#Accounts	25,000	?	?	15,000	15,000	17,000	18,000
#Free Lines*	72	32	0	96	64*1	76	128/300*2
Speed	9.6Kbps	14.4Kbps	2.4Kbps	2.4Kbps	28.8Kbps	14.4Kbps	14.4
Type	Char.	Char.	Char.	Char.	PPP	?	PPP
#Fee	320	130	190	700+50*	96	48	50
Lines				3
Speed	Up to	28.8K	33.6K	28.8K	28.8K	28.8K	28.8K
	33.6
Fac/staff	0	$0.50/hr	$0.40/hr	$10/mo*	$0.75/hr	$0.50/hr	$75/yr
Fee				4
Student Fee	$0.55/hr	$0.50/hr	$0.40/hr	$10/mo*	$0.75/hr	$0.50/hr	Not avail
				4
Type	PPP	PPP	PPP	PPP	PPP	PPP	PPP

Notes:	@	Faculty/Staff count excludes Trades staff and affiliates. Source: COU Report for '95/96
	*	Virtually all institutions offering free services report extreme congestion of these lines
	*1	- 40 free modems for students only, 24 free modems for staff/faculty only
	*2	- 128 modems during the day, 300 overnight, after 6:00 PM
	*3	- Users can pay $15/mo for this group of 50 lines that has a 3:1 contention ratio
	*4	- Will limit monthly hours to 40 for the $10, after which $0.50/hr extra will apply

 

5.0 Eligibility for Access

 

All members of the University of Toronto community should have the opportunity to have dial-in access to the Campus. This includes faculty, staff and registered students whether they be full-time or part-time. Provision should also be made for alumni, and for collaborators whose involvement in University research and academic programs bring value to the University.

 

While observing guidelines for acceptable and appropriate use, all members of this broadly defined University community should be permitted dial-in access at any time of the day from any on-campus or off-campus location within local calling of the UTORdial host. Arrangements for more remote access should continue to be made at the departmental or individual level when circumstances so require.

 

Notwithstanding the desirability of unlimited dial-in access, fiscal limitations require that the provision of dial-in resources must be carefully managed to ensure equitable, efficient access for all members of the University community.

 

 

6.0 Managing Access

 

The importance of dial-in access varies widely among different constituencies and, often, among different individuals within the same constituency. Nonetheless, the number of potential users and current levels of growth suggest that the overall demand could be very large if the barriers to access are negligible or low. The capacity of the system is determined by the number of telephone lines and by bandwidth. Since both are finite, it is likely that usage will have to be limited in one way or another. The question therefore is how to establish those limits in a manner that is fair, workable, sensitive to a range of needs and delivers a reasonable level of service on all fronts.

 

One option is for the costs to be borne centrally, with access provided at no charge to authorized users. This is feasible if the demand can be met without an inordinate strain on the central budget. The service otherwise will be limited to the extent that the University is able or willing to provide and maintain the service. If the service is not adequately managed or maintained, users will have difficulty making a connection and, once connected, they may find that the response of the system is slow. This is rationing by default, in which the costs are borne by the user in terms in time and frustration.

 

The costs of telephone lines, ports and modems are unique to dial-in access. They can be readily determined, and it can be argued that a centrally funded system of considerable size ought to be digestible. If the present capacity were quadrupled, from 320 to 1,280 ports, the capital cost of about $1,000,000 and an annual cost of about $700,000 may be the result. Whatever is deemed reasonable, however, it seems legitimate for overruns to be apportioned in some manner among the users of the modem pool.

 

A somewhat murkier issue concerns the effect of dial-in access on the functioning of the campus network. Many such users will be transferring large quantities of data within the university and, more often than not, to or from remote sites. They therefore will affect the speed of the system for all users, including those with direct connections to the fibre-optic backbone. The number of dial-in ports thus has implications beyond the immediate requirements of access, and the potential cost is appreciable. For example, the cost of maintaining the present 5 Mbps service into the university is $110K annually. A 10 Mbps line would cost about $210K annually. A 45 Mbps link would cost over $300K per year. Since the costs associated with use of the backbone are unrelated to the route of access, any allocation would require that all traffic be monitored and all users charged accordingly. An analogy here is the telephone service, where the cost is charged to the department or other administrative unit, to a research project, or to some other account as appropriate. Such an approach is technically impractical, however, and likely to remain so for some time.

 

Total activity on the network is unaffected by remote access if dial-in users do only what they otherwise would have done from a site at the university. It seems likely, however, that activity is increased. More connect time is potentially available, and the extra hours may be less compromised than hours spent on campus. There also are more users. Direct connections to the campus backbone are limited ultimately by the number of employees, the demands of externally funded research, and the number of stations provided for students; in contrast, the potential number of dial-in users is almost unlimited for practical purposes. A total university population in excess of 60,000 is augmented by paid and unpaid cross-appointees at hospitals and other institutions, collaborators, alumni and a largely undefined group comprising everyone cohabiting with one of the above. Moreover, dial-in access by more than 50,000 full and part-time undergraduates who comprise our largest constituency is likely to mushroom as courses increasingly are structured around electronic links and as more Web-based services are offered to students.

 

It seems self-evident that such numbers have the potential to saturate a modem pool of almost any size and eventually to reduce the performance of the network if access is granted without let or hindrance. At the very least, prudent management argues for an arrangement that encourages efficient and appropriate use of the system with due regard to the different needs of the many different constituencies.

 

 

7.0 Principles and Models

 

Principles for Managing Access in Dial-In Services

 

1.	The central principle is reasonable equity for all users (staff, students, faculty and active retired faculty and staff)
2.	The objective for the Dial-In service is to provide an efficient, cost effective service to the University
3.	Users (faculty and staff) who are not on the University backbone and have no possibility for connection will receive the service "free".
4.	Any future charges for services will be distributed in such a way that individuals will receive some "free" access but access over a certain "point" will be structure that either individuals or departments can cover user costs as required.
5.	Departments who have access to the backbone but have not installed horizontal wiring will receive a period of "grace" ("free" access for given time periods) for one year, after which they will be charged the same as all other users.
6.	There will be special consideration for visiting professors, researchers whose access will be determined at the departmental level on a set time basis (start-end).
7.	There will be a mechanism to determine and evaluate "special" cases and modify access requirements for individuals on a case by case basis.
8.	That any charges for the service will be tied to creating an efficient service and costs recovered will be used first for maintenance and upgrading the service to a set standard. Cost adjustments downward or upward will be based on an evaluation metric to be conducted each year.
9.	All users should have the right under any charging system to pay into that system (add additional tokens) to ensure the appropriate level of service for the needs of individuals or departments. 10. That the fees collected for Dial-In Access are to be directed towards maintenance or the Dial-In service and networking for staff, students and alumni i.e. revenue neutral.

 

Model One

 

All users are charged at the rate of 50 cents per hour and receive at initial set up time one free token.

(A"token" is 20 hours of use, costing $10.00 at the 50 cent rate.)

 

Model Two

 

All users are charged at the rate of 50 cents per hour and receive, one free token per term (i.e., when students are registered (usually two semesters) and faculty (usually three)

 

Model Three

 

All users are charged on a use basis, starting with 10 free hours, 50 cents an hour for 40 hours per month and at a 75 cents per hour over 50 hours per month.

 

It has become evident in the evaluation of these models that it is presently not easy to predict with acceptable accuracy what the impacts of changes to the existing access and charging strategy will be. The university must set up the appropriate evaluation metric to track and evaluate decision making for the Dial-In Service to ensure adequate service, access and cost recovery by users or departments who benefit directly from the service. General agreement was reached by the committee that Model Two was the most desirable, since it was simple to implement, was most equitable and would have the least short term impact on providing resources for the maintenance of the service.

 

The most evident change in each of these models is that staff and faculty accounts would be subject to charges after any initial free tokens were consumed. The decision of how to handle these charges would be made at the departmental or budgetary unit level. The Committee anticipated possible scenarios ranging from the department assuming all excess costs, through determining differing thresholds for various individuals or types of work, up to declaring dial-up access as "optional" and off-loading all costs to the individual.

 

It is very difficult to project the financial impact of the models on local budgets because it is impossible to predict the effect of such a charging scheme on usage. However, a number of calculations may provide a sense of, for example, Model Two's impact. (NB: The average session connect time for a faculty/staff call is currently about 30 minutes.)

 

20min/day	X 5days/wk X 36 wks/yr=	60 hrs - 60 hrs free=	0 hrs @$.50=	$0
30min/day	X 4 days/wk X 40 wks/yr =	80 hrs - 60 hrs free =	20 hrs @$.50 =	$10
30min/day	X 5days/wk X 40 wks/yr =	100 hrs - 60 hrs free =	40 hrs @$.50 =	$20
30min/day	X 6 days/wk X 50 wks/yr =	150 hrs - 60 hrs free =	90 hrs @$.50 =	$45
60min/day	X 6days/wk X 50 wks/yr =	300 hrs - 60 hrs free =	240 hrs @$.50 =	$120
3hrs/day	X 4days/wk X 40 wks/yr =	480 hrs - 60 hrs free =	420 hrs @$.50 =	$210

Anticipating that in most instances a department would chose to subsidize at least some of any additional charges, the committee recognized that concerted efforts should be taken to reduce the administrative burden associated with managing a subsidy strategy at the local level. Creation of user groups to facilitate regular block purchases of extra tokens, departmental access to account balance information, automatic retirement and/or carry forward of token balances, and an easy procedure to "top-up" accounts in emergencies were deemed essential to successful implementation of the plan.

 

Departments should also be aware that requirements for use of e-mail or Internet pages by students in individual courses will necessitate adequate networked computer facilities in-house, around the University or by default they will require students to reach course material via Dial-In access. Departments should consider in cases where adequate network facilities are not available the provision of additional tokens to the relevant student accounts to allow the appropriate access required by such courses. These costs may be considered as part of course delivery or these token costs might be offset by "special" lab or course fees. Each Department should evaluate the impacts on the delivery of networked course information on students and determine an appropriate policy for adequate student network access.

 

 

RECOMMENDATION 1

 

That the University commence charging for all Dial-In Access as of September 1 1997. That students users receive one free token (20 hours of use) per registered term and that faculty and staff receive one free token each term (3). After use of the "free" tokens all users will be required to pay at the rate of $0.50 per hour. Tokens are non-transferable and non-refundable.

 

 

RECOMMENDATION 2

 

That the University continue to make available to alumni dial-In service, commencing with one free token upon account creation (not 3 as is now the case for graduating students) and to charge use at the rate of $1.00 per hour for any additional time. ($1.00 is the rate now currently charged)

 

 

RECOMMENDATION 3

 

That The Academic Advisory Committee of the Computer Management Board (AACCMB) have UTC/CNS evaluate the impacts of changes to the Dial-In Service recommended by this sub-committee. CNS would be charged with the task of monitoring the quality of service, the financial impacts of proposed charges to support the service and to monitor private service provision in relation to the University's own service. CNS would report semi-annually to the AACCMB on Dial-In service provision.

 

 

8.0 Existing Guidelines Regulating Use of University Services

 

At present there are a number of guidelines and use agreements that might be applied to the use of University Managed Dial-In Services. It has become apparent through the course of this report however, that a more universally applicable policy should be clearly stated for all users to ensure proper use of the service for university personnel and university activities only.

 

 

RECOMMENDATION 4

 

That the AACCMB formulate a clear policy on appropriate Dial-In Service Use and circulate this policy to all users of the network.

 

 

9.0 Outsourcing

 

The question of whether the University community is better served by the continuation of the existing in-house dial-in facility or by out-sourcing the entire operation to a commercial provider hinges on four basic points: Cost, Control, Level of Service and Security.

 

9.1 Cost:

 

As of May 1, 1997, the net ongoing costs of providing the current level of UTORdial service are $148,000 per year, with a final loan repayment of $64,902 of capital costs. With the bulk of the capital costs having already been paid, it is only necessary to cover ongoing phone line costs, some support and equipment maintenance from the income of token sales. In fiscal '96/97, student and graduate student use of UTORdial, charged at the rate of $0.55 per hour, generated $264,658 of income. Such a cash flow is more than enough to repay the outstanding capital loan, and cover the ongoing costs. In fact, if, as this sub-committee is recommending, charging is extended to all users of UTORdial, a similar cash flow could be generated at a considerably lower hourly rate, even with the distribution of free tokens, since a much larger user population will be paying.

 

Commercial Internet access providers are profit motivated companies. While many have seemingly low per hour prices or even low flat charges, these are temporary aberrations, as the suppliers compete for market share, while initially operating at a financial loss. The university has no profit motive, and can charge only what is necessary to cover its costs. This will be less than what outside providers need to charge because of the Provost's contributions towards the capital and ongoing costs of the service, which are in effect, a subsidy.

 

9.2 Control:

 

An institution has the maximum degree of control over a service when it provides it itself. With an external provider, one is subject to the vicissitudes of that provider, and moreover, may be only one voice among many when requesting or demanding changes or additional services. A provider is most responsive to such requirements when the institution is the provider.

 

9.3 Level of Service:

 

While there are commercial access providers that offer a high quality service, with few or no busy signals, these are also the higher priced ones, such as Bell's Sympatico. The vendors offering low hourly rates, or even a flat charge service often have all their lines busy for prolonged periods. At York University, where a flat $10 per month is charged, regardless of usage, their modem pool of over 800 modems is reported to be saturated from 9:00 AM to 3:00 AM.. UTORdial, by contrast has been adding phone lines and modems to limit the incidence of busy signals to just a few hours, usually between 9:00 PM and midnight. Once the revised charging policy for all UTORdial users is implemented, there will be a decrease in demand for the service, and busy signals are expected to be a rare occurrence. The performance of the service will be monitored regularly, and additional facilities added as required to avoid the occurrence of busy signals.

 

9.4 Security:

 

While much of the use of UTORdial is by students, there is also considerable use by faculty, staff and administrators, whose data may be of an institutionally confidential nature. With the use of an in-house dial-up access provider, the opportunities for monitoring the flow of data, by 'snooping 'or 'sniffing' are much fewer than if an outside commercial Internet provider were being used. In fact, there has been a publicized instance of an Internet provider's staff monitoring traffic on the dial-up lines they provide.

 

Also, access to the software distribution service UTORdist, as well as some library databases, and administrative documentation is limited to access from U of T network devices. This is done by checking the Internet address of the device. The UTORdial service used by students, faculty and staff is assigned U of T IP numbers to be consistent with this restriction. Users accessing the U of T network via a commercial Internet provider would be prevented from accessing and downloading data from these protected servers, because the IP addresses would not be U of T's.

 

 

RECOMMENDATION 5

 

Given the significant capital investment already made by the University in implementing UTORdial, and hence the relatively low ongoing operating cost, this service should continue to be provided, on a cost recovery basis, until the affordable availability of newer, and faster access technologies, such as Bell's Asymmetric Digital Subscriber Line (ADSL), or the cable companies' "Cable Modems", siphons users away from UTORdial. At that time, UTORdial should be shrunk to keep costs commensurate with income, for those who do not need, or choose to use the newer technologies. To provide equivalent access to University resources through these newer technologies, the security and access concerns mentioned above will need to be resolved, using encryption and authentication techniques. CNS should monitor these issues, including other universities' decisions, and report back to the AACCMB on a semi-annual basis.

 

 

10.0 The Fate of the PACX

 

In October 1996, a subcommittee of the Academic Computing Advisory Committee, charged with performing a Technical Audit of University of Toronto Computing, recommended: "UTC should phase out the current free pool of dial-in lines via the PACX system, switching its phone lines to the UTORdial system according to the current need. PACX service should be discontinued by the end of this academic year and users advised of this as soon as possible. We remain concerned about student access. Please see Recommendation 22." This recommendation was... "We recommend that the Academic Advisory Committee charge a small sub-committee that includes student representatives to explore alternative cost-recovery pricing algorithms for UTORdial."

 

The PACX (Private Automatic Computer eXchange) has been in use for about fifteen years. It represents ancient technology, and is burdened with severe limitations in functionality and speed. It can only support character-based traffic, such as emanates from computer terminals and/or "telnet" sessions from a PC; it cannot support PPP links. Moreover, it can only respond to commands at up to 9,600 bps, precluding the use of modern, 33.6, 28.8 or even 14.4Kbps modems. The benefits of the PACX service are being greatly exceeded by the cost of its phone lines and maintenance. The greatest drawback to the PACX service is its inability to provide authentication. This means that authentication of the user is performed by the system to which a user connects, rather than by the PACX itself. Since there are systems accessible from the PACX that do not require authentication, such as the U of T Library, anyone in Toronto, or anywhere else, for that matter, if they are willing to pay long distance charges, can dial into the U of T PACX, and access the UTL system, even if they have no affiliation with the University. While the Library may view this as desirable, it means that a limited resource can be tied up by people outside of U of T.

 

As the second recommendation above states, this sub-committee is exploring alternative pricing algorithms for UTORdial. A salient feature will be some number of free tokens (hours of use) per semester, after which connect hours would be charged at a lower rate than to-day. This combination should provide an attractive alternative to the PACX. Low usage would remain essentially free, while higher levels of usage would incur some cost. UTORdial will provide greater functionality and speed, and by virtue of being cost-recovery, will be very accessible, with minimal busy signals.

 

 

RECOMMENDATION 6

 

Given that there are summer courses whose students rely on the PACX for access to computing resources, and given the need for some time to implement a revised UTORdial charging policy, with free tokens per semester, we recommend that the PACX be left in service until August 31, 1997, after which it will be discontinued and the phone lines associated with it will be moved to expand the UTORdial service. Notice to this effect should be given to the University community as soon as possible.

 

 

11.0 Appendix 1