By Mark O’Leary, Head of network access
All our members share the problem of providing connectivity to the general public when on their campuses, be it to visit a café or museum, or when attending an Open Day with their children considering where to study. So it may be surprising to some that the Public Wi-Fi DPS, our solution to this for you, is now closing. That’s why I wanted share my thoughts here.
I guess it’s unusual for Jisc or any supplier to take the cover off the sausage machine and show an honest picture of something that didn’t work out entirely as hoped. We did get many things right with the Public Wi-Fi DPS, and it had every chance of taking off, but this time around, and for no single reason, it didn’t do well enough for it to continue in the Jisc portfolio. I’ll tell you a little bit of that story below.
Running a wireless network to accommodate public visitors is largely a solved problem in the technical sense, but problems arise from the users being members of the general public rather than visitors directly affiliated or aligned with organisation’s educational mission. A member’s campus network, as well as the Janet Network it’s connected to, are both considered private networks for legal purposes. This means they avoid a lot of the onerous responsibilities required by public network operators. But a campus Wi-Fi service (and its associated LAN and Wan connections) for the general public must be operated as a Public Electronic Communications Network regulated by Ofcom and follow the obligations under the Communications Act 2003 and other relevant legislation, or risk criminal prosecution. It’s for this reason that the Janet Network connection policy carries a strict prohibition on carrying unencrypted data from a member of the public over a Janet link. See https://repository.jisc.ac.uk/7562/1/janet-network-connection-policy-november-2019.pdf, specifically Clause 28 and associated Note 6. In practice this boils down to the following practical considerations:
- All members of the public must be authenticated to use the service;
- All traffic from members of the public must be carried through an encrypted tunnel across Janet to a public ISP that assumes the legal responsibility for ‘virtual public network’ within that tunnel;
- All traffic from members of the public is identified as originating from the public ISP, not Jisc or the member.
Inside the sausage machine
Three years ago we put in a place a procurement mechanism to help members overcome the challenges of providing Wi-Fi to members of the public. This was the Public Wi-Fi DPS, a form of dynamic procurement framework where Jisc recruited appropriate suppliers of such networks and prequalified them to ensure they fully understood the public/private issue of concern, and had a compliant solution to offer to our members that made use of their Janet connection.
In theory, this was a great approach: our members would save on the cost of public procurement by using our DPS, would be able to access cheaper solutions through using their existing Janet backhaul rather than having to deploy a third-party circuit for the public traffic, and would be putting their requirement before preselected bidders that understood the use case and had the expertise to deliver such a network over their infrastructure.
In practice, we are not extending the DPS contract beyond its initial term, and the service closed on June 30th. So, what went wrong?
First, we were if anything too flexible. Some members would approach the DPS knowing they wanted “a public Wi-Fi network” but without any clear idea of what that phrase meant to them. Would they charge its users? Collect any personal information about them? What volume of traffic did they expect on launch? In 3 years time? Would they cap a user’s bandwidth? The volume of data downloaded? How regularly a given user used it? It often took a couple of hours of meetings, with gaps to allow the team to go away and consult colleagues to answer some of these questions before an invitation to tender document could be produced to put before the suppliers within the DPS. We had designed the DPS to be flexible enough to help a member find a bidder for any style of bespoke guest network they could think of, when with hindsight some members at least would have welcomed a ‘take it or leave it’ fixed option at a fixed price.
Another issue was getting a reasonable number of bids from the DPS suppliers for it to be a competitive environment. We don’t know if these public solutions were perceived to be too small to be commercially viable or too onerous in the legal sense, or just whether sales opportunities not arising from their own sales team were given lower priority, but what we found was that some invitations to tender received only one or two bids and a few didn’t get any bids at all. Given the effort that was going in to defining the requirement, this was disappointing for the members looking for a solution, and for us in recommending an approach to them that didn’t deliver.
The final problem was that very few members were actually making use of the procurement service. Whether through having an existing in-house solution, being locked into a long term contract, or being marketed to directly outside of the DPS, the majority of general public guest networks on education campuses deployed in the last three years were not procured via the Jisc route.
So, as we approached this Summer, the future of the Public WiFi DPS was in doubt. Then, ironically, we had a sudden burst of activity, with a number of Universities expressing interest. I did initially indicate to some of these that we were likely to extend the DPS to accommodate this late flurry, but a contractual issue meant that if we extended at all it would have to be for a minimum of two years, and I was rightly challenged by colleagues that extending a poorly-performing service for this long would not be in the best interests of our membership overall. Ultimately the decision was taken to close it to new business at the end of its existing contract.
What are we going to offer instead? We are going to reflect on the performance of this option, and talk to some of you about what worked and didn’t work for you around the DPS before we dive into the next solution. I’ve already sketched one option, which might be to partner with a public ISP to deliver a single much simpler white label product that members could choose to buy or not; we might equally take this on as a development project and design from scratch something sector-specific more along the lines of eVA, or any of a number of other different models we might pursue. I’d welcome input from you all as to what you’d like to see.