To the University Community
Penn's new model for computing services weaves computing into the academic and administrative fabric of the University. The model makes sense in Penn's environment and positions the University to take advantage of information technology in new and exciting ways. We invite your guidance and participation in the continuing process of making computing easier and more cost-effective for everyone who uses it.
--Stanley Chodorow, Provost, and John Fry, Executive Vice President
Abstract
Computing now touches everyone at Penn. Those who use and those who provide computing services recognize that Penn's structures for sup port can be improved. The model presented here was developed by a campus-wide task force in the fall of 1995 and vetted across the University. Pilot projects are underway to test and refine the model.
The model clarifies a division of labor at Penn. Primary computing services will be provided close to the user by schools and units, while core infrastructure and second-tier support will be delivered by the central computing group, confederations, or outside vendors. Two strategies help shape secondary services: networking as a regulated public utility and service bureaus where markets exist. The model also offers a potentially powerful way for Penn to take action at the University level; a few cross-cutting processes will be funded directly and managed across traditional organizational boundaries.
Contents
For more information
Contact Linda May for more information (may@isc.upenn.edu; 215-898-0005).
The model doesn't claim to do everything. It doesn't ignore history. It is a way of doing business that gives members of the community the chance to make Penn better and exposes each of us to the costs of bad decisions and the benefits of good ones.
What Problem Are We Trying to Solve?
Those who use and those who provide computing services recognize that Penn's structures for support can be improved. Some things are needlessly complicated. People don't always know where to go for help. It's hard to tell what things cost. Changing things won't be easydemand is soaring, technology changes relentlessly, and Penn is a very complex place. But we have accepted the challenge to make computing work better at Penn.
Each of Penn's twelve schools supports the technology needs of its faculty and students in different waysand the principle of Responsibility Center Management requires us to expect the schools to pay their own way. Some, but not all, of Penn's administrative divisions have their own computing staffs. The central computing group, Information Systems and Computing, provides services that range from essential infrastructure best managed University-wide (the network, for example, or payroll) to frontline user support. Central/peripheral tensions are played out at several levels: from center/school to school/academic department. The Library is caught in the middle of technology decisions made by ISC and the schools. And everywhere people need more and better support. In short, we have Responsibility Center Management in principle, but a messy situation in practice.
Principles
For computing to be applied strategically at Penn, it must be easier and more cost-effective for the people who use it. To this end, the task force
took Responsibility Center Management as a framework and tried to unite responsibility and authority where they have grown apart. At the same time,
we tried to focus Penn's actions at the University level. And we tried to create incentives to integrate computing decisions into the core decisions of
the University.
The new model is based on the following principles:
Basic elements of the new model are described here. More detail is found in four appendices. The project's World Wide Web site can be seen at http://www.upenn.edu/restruct.
The User. The computer user is at the center of our model. Each person ideally has a local computing "home" and takes all computing questions there. Beyond this circle of primary support are expanding circles of secondary servicesprovided by ISC, by confederations, or by outside vendors. But the map of services is irrelevant to the recipient: the primary support person navigates that landscape.
Primary Services. In the model, schools and administrative divisions are responsible for their own primary computing support. This includes frontline customer support (including the desktop computer and its relationship to the network) and support of local academic and administrative systems, services (including local-area networking), and innovations. Units can provide primary support themselves or buy it from other schools, from ISC, or from outside Penn. The task force urges that guidelines for basic primary support levels be set and that Penn institutionalize ways to keep these levels moving up.
The model makes primary support local so that decisions are based on the real cost of service. Primary support providers can do a good job of telling users what things cost and helping them make responsible choices. The model seeks to end current incentives that lead people to demand unlimited services. (Allocated-cost service is "free"because already paid forthe reasoning goes, so why not ask for more?) The model also seeks to end unfunded mandates at every level. Schools may reasonably fear that burdens will shift to them as ISC stops offering primary support as an allocated -cost service to the general community. This reflects, however, the extent to which ISC has been the recipient of unfunded mandates in the past, a practice the task force recommends ending as efficiently as possible.
While frontline support for faculty, students, and staff is the responsibility of Penn's schools and business units, their ways of delivering and funding that support will vary widely. For many faculty, for example, the department is the natural computing "home"yet economies of scale are needed. The task force recommends that schools explore departmental coalitions and other affinity groupings based on location, discipline, or type of computing. For undergraduates, the task force recommends moving over time to residence-based support, building on broader efforts to restructure student services across Penn. Good models for residential support exist at Stanford and Northwestern. How fast Penn could move in this direction and how to support undergraduates in the meantimeare under discussion.
Secondary Services. Secondary services undergird primary support and make the whole greater than the sum of the parts. The task force calls for a more focused set of such services: core administrative systems, networking, data administration and information security, second-tier support for computing organizations around campus, standards and architecture, site licensing, and communication at the enterprise level. Penn's central computing group, ISC, will concentrate on these services. A few may be candidates for delivery by confederations, individual schools, or outside vendors. ISC will review each of the services it now provides, eliminating some and focusing more heavily on others. While most of these secondary services will continue to be funded by allocated costs, Penn will move over time, as indicated below, to market-based structures where they make sense.
A vital central function that aids confederation and contains costs is the negotiation of standards across Penn. The task force stresses that standards succeed at Penn only when they are worked out by the community itself. The many and sometimes hidden costs of compliance with standards otherwise become another kind of unfunded mandate. Incentives to adopt standards will be built into the support structure.
Two new strategies help shape secondary services:
Process Teams to Focus University Action. The model offers a potentially powerful way for Penn to take action at the University level. A few cross-cutting processes will be funded directly and managed across traditional organizational boundaries. For the moment, Penn will concentrate on two or three high priority processes such as academic innovation and student services. These processes can be considered "institutional bets" with high potential payback. As political implications of the process perspective are worked out, more of Penn's daily life may be organized and funded along process lines.
The task force can't guarantee that process teams will transform the institution, but we can say that the innovations they achieve will not distort the system, create unfunded mandates, or break the bank. Processes will be funded partly at the University level and partly as units bring people, dollars, or facilities to the table. The high visibility of process teams can also draw outside funds. Process teams will buy services from existing organizations, strengthening the evolving market economy of the new model.
Process teams are confederacies. The task force on restructuring computing has already been this kind of confederated team, drawn from across the University to do the University's business together. To work well, such groups need to bring the real interests of Penn's units to a common project. They will be created not by any special mechanism, but by the responsible decision makers of the institution.
Governance. Computing has become essential to almost all fields of research and instruction and to Penn's administrative lifeand Penn is investing heavily. The task force calls urgently on the leaders of Penn's schools and units to integrate computing decisions into the regular decisions of the University. Computing is no longer just a technical specialty, but a strategic advantage. We also urge the Provost and Executive Vice President to explore the feasibility and desirability of convening a chartered group to consider issues of campus-wide importance to Penn.
Costs. Cost-effectiveness, targeted investment, and giving units more control over their costs are aims of the model. In an area where Penn's investment is sure to expand, our sponsors want to see money saved in some areas and reinvested in others. For example, as ISC reduces the number of things it does for allocated costs and cuts costs in other ways, funds can be returned to the Provost's budget. The task force strongly urges that these savings be spent on forward-looking computing activities. Process teams are a prime target for these funds.
We freely and frankly say that we cannot tell whether this model will be seen by individual units as costing them more or less. Rather, in an environment of exploding demand, the model will give units more control over their costs. The model tries to unite responsibility and authority where they have grown apart, to reveal real costs where they have become obscured, and to return choice to purchasers where it has been eroded.
All in all, the model clarifies the division of labor under Responsibility Center Management. Schools (and business units) are responsible for their own primary computing support. They can provide it themselves or buy it from others. The center concentrates on secondary services. Standards help tie the structure together. The model encourages confederation for the common good even as it values organizational self-reliance; process teams, for example, focus action at the University level. The model calls on Penn's leaders to integrate computing decisions into other core decisions of the University. And it aims to give units more control over their costs.
More details of the model can be found in four appendices:
The task force has completed the design phase of its work. The Provost and Executive Vice President have appointed a much smaller Implementation Steering Group (Appendix II) to stimulate and oversee pilot testing and transition to the new model and to further develop funding structures. The sponsors and steering group continue to consult with leaders of Penn's units and negotiate ways of applying the new model. For example, the question of primary support for all will need careful examination in virtually every unit. ISC is prepared for a transition of 18 months (from January 1996) to withdraw from providing primary support as an allocated-cost service to the general community (though contract or other arrangements can be negotiated with ISC).
Some of the analytical work of the original task force continues, notably in two teams (Appendix II) working on funding (how to pay for what we do, now and in the future) and benchmarking (what are best practices elsewhere for questions that arise here).
The following pilots are underway to test the new model:
The Steering Group will guide and integrate these pilots. It will draw lessons from the pilots and revise and renegotiate the model in light of lessons learned. With the Penn community, it will design responsible funding structures and lay groundwork for transition to the new model.
ISC and other Penn computing organizations are restructuring in line with the new model. ISC, for example, is rethinking roles and responsibilities, sharpening its focus on enterprise services, and restructuring to provide other services on a direct-charge basis or as regulated utilities. Other units at Pennthe School of Arts and Sciences is a notable exampleare likewise rethinking the services they provide and the ways they provide them.
Appendix I:
In the fall of 1995, Penn's Provost and Executive Vice President appointed a University-wide task force to make computing easier and more cost-effective for those who use it. Our charge was to design a new structure for organizing, staffing, and funding computing services across Penn.
Task Force to Restructure Computing Services Across Penn
Appendix II:
In the spring of 1996, the Provost and Executive Vice President named an Implementation Steering Group to guide pilot testing of Penn's
new model for organizing, staffing and funding computing services across the University.
Implementation Steering Group
Appendix III:
Seven pilots are testing Penn's new model for organizing, staffing, and funding computing services across the University. Team leaders are noted below.
Leaders of pilot teams
Appendix IV:
In the new model, schools and administrative divisions are responsible for their own primary computing supportaffirming the principle of
Responsibility Center Management. They can provide it themselves or buy it from other schools, from ISC, or from outside Penn. In general, primary
support will encompass frontline customer support (including the desktop computer and its relationship to the network) as well as local academic and
administrative systems, services (including local area networking), and innovations. The task force urges that guidelines for basic primary support levels
be set and that Penn institutionalize ways to keep these levels moving up.
Primary Services
Frontline support will be backed up by secondary services provided by the center, by other units, or by outside vendors. For the computer user, the primary support provider is the link to those services.
In practice, the distinction of primary and secondary support is not a dichotomy, but a continuum of services appropriately sited. For example, an individual laboratory might support its own local area network (LAN), the academic department that sponsors the laboratory might handle LAN upgrades, a set of departments might share a LAN expert, and the central computing group might make available a network engineer. In our model, the primary support personnot the usernavigates these complexities.
Principles. Because primary support is key to our model, we have described the ideal, our target, in some detail below.
Campus units that choose not to provide primary support to their members should expect to pay a premium for their members' access to second and thirdtier support elsewhere in the institution. This is fair because the unsupported user is likely to bring small problems to expensive places and may not have exhausted cheaper remedies.
Mechanisms are in place to refer problems that a primary provider alone cannot solve. Sources of second and thirdtier support are clear, and means of access are well defined.
Onsite support for remote users: Remote users who require onsite support that cannot be delivered by the primary provider should be referred
elsewhere within Penn or to preferred commercial vendors. Mechanisms are needed to evaluate and monitor those arrangements and negotiate the best
prices. Penn's remote users range from faculty and staff on leave or stationed elsewhere to those who sometimes travel or work from home.
Business consulting: Primary support providers are expected to integrate needs, resources, technology, data, etc. into a coherent support
environment. Some problems, however, will require analysis that primary providers may lack the time or experience to perform. The task force suggests
that a market may exist at Penn for a service bureau that helps people analyze business problems and assess possible solutions.
Informal support: Staff serving other functions can be very effective at delivering primary computing support if secondary support is in line to back
them up. Staff already in place can do triage and "first aid" if they know how and when to pass people to the next level. This "Hey, Joe!" support is
common at Penn and needs to be acknowledged in job descriptions.
While most of these services will continue to be funded by allocated costs, Penn will move over time to market-based structures where they make sense.
ISC will give the Penn community a periodic accounting of what it receives for allocated costs.
Two new strategies for delivering secondary services are described below.
Network as a Regulated Utility. Penn's network will be run as a utilitywith service-level agreements, campus-wide standards, and a "public utility
commission" (PUC), or governing board, to keep it responsive and competitive. The PUC will approve tariffs, service levels, and standards; seek the
input of the community; and participate with others in strategic planning for the network. The PUC governing board will be drawn from the units of
the University and will include network-intensive researchers. As a utility, the network will be funded by a mixture of allocated and direct charges,
with specific funding strategies to be taken up by the PUC.
Task force discussions have focused on PennNet and the Penn Video Network, both operated by ISC, with a view to making these services more
like Telecommunications' telephone service, which operates like a traditional utility (but without a PUC). Potential extensions of the model include
a common PUC for all three services and a common management structure for all three.
Currently, the Penn Video Network approaches the utility model, while PennNet and its related services are more mixed. In general, the PennNet core
(backbone routers, inter-building fiber, etc.) comes closest to a traditional utility. Central PennNet services (authentication, directory, News, Web, list
servers, etc.) operate somewhat like a utility. And satellite closets, station wiring, LAN servers and services, and email are subject to varying practices
and interpretations. The Network Policy Committee has functioned, to some degree, as a PUC (with annual review of proposed rates, participation in
network architecture planning, etc.) but has not had the governing authority that a PUC would have.
Market-based Service Bureaus. The model moves Penn toward a market economy in some areas, to control costs and encourage a customer focus.
Where it makes sense, "service bureaus," or small businesses, will sell products or services or hire out individual professionals to local projects or longer
term assignments. Wharton Reprographics and ISC's support-on-site program are examples of service bureaus that already exist at Penn. Any unit is
invited to set up a service bureau. In practice, most will likely be run by the center.
Service-level agreements will define offerings, scope, cost, and performance measures. Primary providers will act as intermediaries or partners in
such arrangements. With a healthy set of service bureaus in place, local units might see their own staffing needs shrink.
In their pure form, service bureaus compete in the open market. Strategic concerns might warrant an allocated component for some service bureaus.
Start up might be funded by University seed money or loans, prepayment by key clients, or pooling of funds. Transition strategies (for ISC units in
transformation) might involve a tapering off of allocated fundsor perhaps a giveback of allocated funds with a promise to spend them with the service
bureau for a period of time. The task force recognizes the difficulty of moving to a market economy. It knows that transition will take time and that
not every new business will succeed.
Computing is just one element of these processes. The academic innovation process, for example, draws together schools, the Library, the Division
of University Life, the Classroom Committee, and the lab managers' interest group, among many others. In practice, major processes encompass smaller
ones. Academic innovation encompasses smaller processes such as classroom renovation and the introduction of software such as Maple into the
curriculum.
Process teams are confederacies that link authority, responsibility, and funding. Leadership comes from wherever appropriate. Processes have
"owners" who make sure that the work of the process doesn't fall through the cracks between traditional organizational units.
Process teams have a life cycle. The initial
political work is done by the process team; members need authority to commit resources and make
decisions. As political issues are settled and the process becomes more routine, activities can be handed off in different ways. Labs, for example, might
start out within a process team and be handed off to a lab managers special interest group. Some process teams may evolve into new types of formal
organizations.
An opportunity to be explored is how the process perspective (and underlying machinery) can improve center/local interfacesand more
importantly, how the new perspective can influence the core activities of both central and local organizations.
A funding framework for the new model is being designed by the Implementation Steering Group, the pilot teams, and a subgroup of financial
specialists. The framework includes principles, operating guidelines, and specific mechanisms and structures. We are assembling a framework pilot
by pilot in order to construct a campus-wide strategyif that proves to be appropriatefrom the ground up. Eyes on the horizon, we are concerned
with "piloting the funding," not just with funding the pilots.
As part of this effort, and in keeping with the model's emphasis on local control, we are documenting basic building blocks of costs and funding--for example, support ratios, salaries for different kinds of computing staff, and costs of equipping a staff member. Penn's units can combine these
building blocks in ways that make sense for the unit. Or the unit might look for ways to trade off one type of cost against another: substituting capital
for labor, for example, or transforming existing staff positions into computing positions.
The funding framework, like the entire model, rests on two basic assumptions about computing. First, computing isn't something that happens
"somewhere else." It is inseparable from Penn's mission and touches every member of the University community. Second, funding for computing is
everyone's business. The computing infrastructure is now part of the campus infrastructure and should be factored into fundraising strategies and
budgetary planning.
Principles. All of this activity is guided by three funding principles, implied by the model's organizational elements:
Operating Guidelines. Guidelines such as the following inform the evolving funding framework:
Guiding Questions. Questions such as the ones below are helping us develop specific funding models and mechanisms. In practice, for example,
funding mechanisms act as incentives for behavior. Supporting a service wholly with allocated costs, for instance, may encourage customers to think
of it as "free," and stimulate demand, while supporting it wholly with direct charges may limit use. Penn will want some services to be perceived as
free, or nearly so, and others as carrying a cost.
A New Model for Computing Services Across Penn
Please send your comments
by June 7, 1996, to:
Linda May
(215) 898-0005
e-mail: may@isc.upenn.edu
Almanac Supplement
Almanac
Volume 42 Number 31
Return to index for this issue.
Appendix V:
The model recognizes that some types of computing services make the whole greater than the sum of its parts. The task force calls for a more focused
set of such services, outlined below. ISC will concentrate on these secondary services. A few are good candidates for delivery by confederations,
individual schools, or outside vendors. ISC will review the services it now provides, eliminating some and devoting more energy to others. With the
community, ISC will perform periodic "sunset" reviews of services, processes, and organizations.
Secondary Services
Appendix VI:
The model seeks ways to focus strategic actions at the University level. One potentially powerful strategy is the direct funding of broad processes
that cross traditional organizational boundaries. These processes will be funded partly at the University level and partly as participants bring people,
dollars, or facilities to the table. For the moment, Penn will concentrate on a few high-priority processes such as academic innovation or student services.
Over time, more of Penn's daily life may be organized and funded along process lines.
Cross-cutting Processes
Appendix VII:
Cost-effectiveness, targeted investment, and giving units more control are aims of the model. With responsibility for primary support squarely at
the local level, units have more control over what they spendand greater control brings pressing decisions about the funding and staffing of services
offered their members.
Developing Funding Structures
Director of Planning Information Systems and Computing
Suite 230-A
3401 Walnut Street
Philadelphia PA 10104-6228
May 7, 1996
Return to Almanac's home page.