Docstoc

Making Strategic Decisions about Bandwidth

Document Sample
Making Strategic Decisions about Bandwidth Powered By Docstoc
					Tertiary Education Network

Making Strategic Decisions about Bandwidth
A discussion document

Version 0.2 23 January 2007

Summary: Defensive bandwidth management fails to connect tactics to institutional objectives. At best it eliminates abuse but does not secure any positive outcomes. Strategic bandwidth management seeks to reserve scarce bandwidth for core institutional purposes. Background. The Internet revolution has permanently and fundamentally altered the way in which universities function. At a conceptual level, they remain (or ought to remain) what they have always been, but the way in which information is produced, shared and consumed is now so heavily intermediated by information technology that a university depends utterly on the quality of its connections to both the commodity Internet and the global research network. Whether a researcher is in Stockholm or in Kinshasa is, at one level, irrelevant: the same technology imperative applies, to the extent that the applications, the clients, the protocols and the operating systems are substantially the same. But the available bandwidth is not the same. The regulatory and competitive environments, the availability of terrestrial infrastructure, the distance from the global network cores and the cost and quality of bandwidth are all very different. The result is a dramatic disparity in bandwidth provisioning between the developed and developing worlds. Typical institutional bandwidth in the former case is in excess of 50 kbps per client; in the latter case it is often 1% of that. Remedying this by buying more bandwidth is usually out of the question, since the cost is often two orders of magnitude greater than in the northern metropoles. A developing country university has to manage its bandwidth in a completely different way from a developed country institution. Defensive and Strategic Management. Bandwidth management is a function that almost invariably falls to the IT department, and it is almost invariably handled defensively. Network administrators strive to improve the technical efficiency of circuits, and (sometimes) to allocate available bandwidth as fairly as possible. They respond to abuse reactively, by closing loopholes. But for the most part they don’t step back from the problem and ask questions about the strategic management of bandwidth. By “strategic management” I mean an approach that connects the management of the bandwidth as tightly as possible to institutional objectives, and is willing to sacrifice the lesser good for the greater. There is good reason why network administrators don’t do this: it is simply beyond their purview, since the resource that they are managing is a collective good for the entire institution, but their domain of responsibility is technical rather than institutional. Furthermore, they occasionally burn their fingers when they venture into the realm of qualitative management. A good example is the use of content filters. These are fairly crude instruments that block certain traffic on the basis of its properties – such as keywords, URLs, or IP numbers. Some content is deemed bad (because it is manifestly intended for private gratification) and is therefore filtered out. But such filters are very inefficient, and their application by the IT department often elicits a storm of protest: the IT department is accused of being self-appointed custodians of public morals, or of being an enemy of academic freedom. For good reason, therefore, network administrators confine themselves to quantitative management. But such an approach cannot even begin to address the important strategic questions.

Page 1 of 4

As a way of thinking about such strategic questions, let us imagine five different kinds of traffic. Technically they are all the same, in the sense that they are delivered by the some protocols to the same applications. But they are qualitatively different, not because of their manifest properties but because of their value to the institution. The five kinds of traffic, their values, and the typical institutional management response are summarised in table 1.

Table 1. Typical institutional response to five classes of traffic. Example Hacking Value High negative Explanation Potential for damage to institutional reputation, or even legal liability Wasteful of resources Improves user connectedness to the digital world but no direct academic benefit Improves user connectedness to the scholarly world but no high-quality benefit Direct positive consequence for teaching and research Typical response Firewalling and policy

Pornography

Low negative

Policy

General internet

Neutral

None

Academic blog

Low positive

None

Electronic journal

High positive

None

In this scheme of things an attempt is made to eliminate the negative value items. Some of this can be done by technical measures (generally on the firewall) and some by means of policy – which, if rigorously enforced, can dramatically reduce traffic in pornography and pirated digital content. But no attempt is made to distinguish among the other three classes of traffic. The neutral- or low-value items enjoy the same access as the high-value items: they compete on equal terms for the available bandwidth. Since the overall demand far exceeds the supply, some requests fail. Since the proportion of general internet requests far exceeds the number of academic requests, far more of the former will succeed. And since the academic requests are generally for objects of a relatively large size – a journal article, for example, is generally delivered as a PDF file exceeding a megabyte in size – they are especially vulnerable to TCP session failure, which is what always accompanies bandwidth congestion. A management regime of the kind pictured in this table will in effect penalise accesses to digital scholarly resources. A strategic approach to bandwidth management must seek to maximise the traffic in the fifth class. For a developing country university, digital scholarly resources are the single greatest opportunity that is offered by the Internet revolution. They are available, through donor and publisher initiatives, at relatively low cost, both compared to costs in the developed world and to the cost of physical journals. Sometimes they are free. Yet a tremendous opportunity is cast aside because bandwidth is managed defensively rather than strategically. Defensive management means that general Internet access – (Google, CNN, howstuffworks.com, and millions of other sites) enjoys the same privilege as www.jstor.org. And within the framework of a defensive management regime, there is no alternative to this. The neutral-value items

Page 2 of 4

can’t be managed downwards, because there is no mechanism, either technical or policybased, for doing so. A strategic approach, by contrast, would not seek to manage some classes of traffic downwards. Rather, it would ensure that some classes of traffic have privileged access. Given a resource constraint, the most effective approach is sometimes to channel the resource towards the most valuable uses rather than to channel it away from the least valuable. In the context of bandwidth management there are many ways of achieving this, but the simplest and the cheapest is to segment the bandwidth among the elements of a cache array, and direct some requests through a priority cache and leave others to take their chances in a general-purpose cache. In Table 2, the three “none” cells now have positive content: Table two. A strategic approach to five classes of traffic. Example Hacking Value High negative Explanation Potential for damage to institutional reputation, or even legal liability Abusive and wasteful of resources Improves user connectedness to the digital world but no direct academic benefit Improves user connectedness to the scholarly world but no high-quality benefit Direct positive consequence for teaching and research Typical response Firewalling and policy

Pornography

Low negative

Policy

General internet

Neutral

Benign neglect

Academic blog

Low positive

Priority access on academic request

Electronic journal

High positive

Priority access managed by the Library

In this regime, some bandwidth – perhaps even most – is consciously and intentionally reserved for resources that, by common agreement, have the greatest institutional benefit. No disability is imposed on the class of general Internet resources; requests for these resources take their chances in a competitive arena with no special attempt either to filter or to prioritise them, apart from ensuring technical efficiency in processing them. But any resource can be promoted, given sufficient academic demand, to a more privileged space. A regime of this kind entails an institutional acknowledgement that bandwidth scarcity needs an approach that is qualitatively different from that which is typical of developed country contexts, where these kinds of constraint almost never apply. Resource abundance requires no strategic decisions; resource scarcity does. Isn’t this an acknowledgement of defeat? At one level, it could be argued that this is true: certainly it is an acknowledgement that the circumstances of higher education in much of the developing world are simply different from those in the metropolitan regions. But such acknowledgement is damaging only to the extent that it means abandoning the objective of substantive parity, and that objective should certainly not be abandoned. Universities in developing countries should vigorously pursue such parity, through the formation of National Research and Education Networks and the quest for access to terrestrial connectivity at a realistic price. Until those objectives are attained, however, there is nothing to be gained by

Page 3 of 4

pretending that exigent disabilities are not present. They are, and they impose an inescapable obligation to make strategic decisions about the limited bandwidth that is available. Making this happen. Although a regime of this kind is implemented at the technical level, it has to be underpinned by an institutional consensus. This involves (a) a recognition that a strategic approach is necessary; (b) an agreement among the stakeholders – institutional leadership, the academic community, the library and the IT department – on the shape of that approach; (c) a technical design and (d) a set of processes that serve the ongoing operational requirements of such an approach. TENET can help with facilitating conversations among stakeholders and with training for technical solutions. This is a discussion document. Responses to dbg@tenet.ac.za are sought.

Page 4 of 4


				
DOCUMENT INFO
Shared By:
Stats:
views:19
posted:12/15/2009
language:English
pages:4
Description: Making Strategic Decisions about Bandwidth