SUNET C will consist of 7000 kilometres of dark fibre, separated into a number of redundant rings, covering the whole of Sweden. To procure all the dark fibre needed, we of course had to get it from some supplier. For this, we had to follow the strict rules for public procurement stated in the Law on Public Procurement (Lagen om offentlig upphandling, LOU), send invitations to tender, weigh the pros and cons, and so on.
What pops up in your mind when you read “public procurement”? Probably the never ending row of appeals from various vendors who consider themselves passed over, or some who quoted unreasonably low prices, whereupon other vendors, in turn, quabble fiercely. But alas, no. Thanks to SUNET’s great technological professionalism this round of procurement just ran smoothly and quietly. Everyone involved was happy and no one complained. Quite simply, an ideal paragon for others to follow.
We must admit we’re slightly chocked considering today’s tough procurement climate, where virtually all sorts of dirty tricks are allowed. The person responsible for the business transaction was Fredrik Person Jonhed, manager of procurement at Karlstad University and SUNET. All right Fredrik, explain yourself!
– It’s all about dialogue, risk evaluation and procurement skills. The Law on Public Procurement allows for several methods of procurement. Our process was very complex and we had to have an ongoing dialogue with the suppliers, so we choose the “competitive dialogue” method. The way this is done is that you begin by setting up a number of supplier parameters, such as size, capacity and references, whereupon you publish your interest. We got six tenders and selected four of them for the second round: Tele2, TeliaSonera, Triangelbolaget and TDC. They were all invited to a dialogue during the autumn of 2014.
The invited companies were given a preliminary, suggested invitation to tender that they could study and use to prepare their answers, and most of all, to indicate if there were any problems with the questions. The idea was to get feedback from the market on what was reasonable and to find out if we had set up any unreasonable requirements. We wanted to be very clear about what was important in their answers, and most of all, how they answered.
Let’s say we had some requirement and some supplier choose to say “no”. They would then have to be prepared to persuade us that we had set up a bad or impossible requirement. On top of that we were forced to make another decision. If the supplier answered “yes” to our questions, we hade to make sure that they were actually able deliver the solution, and if so, how. They had to go to great lengths to prove that they were actually able to do it.
There were lots of arguments for and against. Initially we didn’t want suspended fibre, that is, fibre hung in power line pylons, but rather argued for buried fibre. The suppliers indicated there were many advantages of suspended fibre. We had to change our preferences. The Västerås Incident was proof enough that buried fibre isn’t entirely safe. It’s also harder to repair. It turned out that the risk of failure of a fibre hung from pylons was smaller. We ended up ordering both kinds.
Another, similar question was our requirement to get the KML files, for displaying the fibres’ exact location in Google Earth. Some suppliers were not willing to disclose that information, and to retain the suppliers in the ongoing process we had to let this requirement fall and be satisfied with simple maps.
Those were examples of compromises we were forced to make. After all, it doesn’t matter if the cat is black or white as long as it catches mice, Fredrik points out.
And then Fredrik discloses an important insight into the darkest abyss of the procurement process, that probably very few in his position are aware of:
– A nice SLA is of no value when the network actually goes down. We may be able to sue the supplier for non-delivery, but that’s not enough. We don’t want problems. The SLA may be sitting in a folder somewhere, but we don’t want to retrieve it. Instead we want a supplier we can trust. It doesn’t matter if we are able to claim a few hundred thousand Swedish crowns because the network has been down too long. We are not interested in the money. Having the Max IV laboratory or the Onsala Radio Telescope left without Internet for a week is completely intolerable. The stuff we buy, must work. No exceptions.
Disrupting thoughts, indeed. We progress.
– After the first get-together we went back hone and wrote another questionnaire, this time including the suppliers’ feedback. There was no great difference to the first one, but now we asked them to answer as if it had been a procurement.
We also informed them that we would grade and evaluate the answers for risk, grading from zero (disqualification) to 6, the reason being that we simply didn’t want any bad answers. Fuzzy answers tend to lead to trouble, or preventing a good solution from being properly presented, and as such may be dropped in error. Also, this increases project risk.
After we got all the answers, we wrote a final invitation to tender. At this stage, none of the suppliers could argue that they hadn’t understood our objectives. We knew that all four of them were able to deliver the technical solution in an appropriate way, so we choose to put more weight on the cost side. This meant that in the final judgement quality would be weighted at 30% and cost was weighted at 70%. Still, we maintained and made it very clear that the most economical alternative may not necessarily be the cheapest one.
The Christmas present
All the answers came back at Christmas 2014. The answers were more or less known already, and Tele2 came out at the top. We waited for the required appeal time, but no one appealed, and so the deal was done. We obviously did the right thing, by clearly informing about what we wanted, evaluated all the questions the way we said we would, didn’t add any new requirements at the last minute, and we treated all suppliers equally.
Party with cake
The agreement was signet at April 14, 2015 and we had a huge cake at SUNET’s office in Stockholm
This is the moment when SUNET’s chairman of the board Stefan Bengtsson shook hands with the Tele2 managing director Thomas Ekman, after the deal was done. Image: Tele2.
You might find it interesting to know that the agreement is a mere ten pages. This is unusual for a Swedish hundred-million-class agreement. 40 pages would be the norm. This raised some eyebrows, but we think it’s advantageous. Why include a lot of clauses from the Sale of Goods Act and the Contracts Act in the agreement, when they are clearly stated in the law already? Anyway, all the invitations and tenders are enclosed. Our aim is to use the agreement primarily to solve everyday situations, both within SUNET as well as with our partner Tele2. The whole process started at midsummer’s eve 2014 and ended happily in April 2015.
The network we purchased is partitioned into a backbone network, access networks and some smaller extra Ethernet links, as well as the placement, that is, premises where to put the equipment. Everything was bought in one go, but pending our final solution we are not sure about the number of In-Line Amplifier nodes. This will be an additional cost for SUNET, and this had to be taken into account when selecting vendors. The backbone and the premises represent the large costs. The access networks, that is, the city networks, already exists, hence they are but a small part of the whole.
– The cost is split in two. We wanted to buy the right to use the backbone for 15 years without the supplier having any chance of taking it back (IRU, Indefeasible Right of Use), rather than just renting it. Also, we wanted to pay off that cost in advance to get it into the investment budget. After that, we could start planning our equipment procurement, and include the cost of a generational update of the In-Line Amplifiers.
Add to that a monthly cost for running and maintenance, which has been included in the published total price. According to the Service Level Agreement, the monthly fee will be reduced in case of a communication breakdown. The reduction amounts to 2 % of the monthly fee per commenced 90 minutes of downtime exceeding the initial 90 minutes of downtime.
It is up to SUNET to procure equipment for routing from the backbone to the access networks. Basically, we get some loose fibre ends in the points-of-presence that SUNET’s staff will have to contact, connect and test. The same goes for the forest huts housing the In-Line Amplifiers between the cities. Tele2 maintains the huts, whereas SUNET furnishes the equipment inside. SUNET also pays for the electricity powering the huts.
How will Tele2 make sure no two fibres will ever get closer than ten metres from each other, this time (remember the Västerås Incident)?
– Tele2 has promised us this will not happen, and we are looking closely at their maps to make sure. Regarding the fatal Västerås Incident in OptoSunet, we simple hadn’t got what we paid for. We might be generous and overlook a smaller separation distance in some special cases, such as inside buildings, or at electrical substations where no one is allowed in with a digger. As part of the agreement, SUNET’s costs will be reimbursed for non-delivered diversity in the case of another Västerås-type incident.
SUNET has demonstrated a skill that very few in the business possess: the power to stay friends with all vendors despite the prospect of one of them getting a very lucrative contract. It’s all about information. Conducting this purchase as an open procurement in accordance with the Law on Public Procurement, that is, without continuous discussions with the suppliers, would have been impossible. Despite this, open procurements are by far the most common today. Also, the insight that the SLA is no more than a worthless piece of paper when the network goes down, will strike many others like bomb.
– We have focussed on eliminating projects risks by putting proper questions to the suppliers, and properly evaluated the answers. We want to be sure the suppliers can actually manage to deliver what they promise, Fredrik concludes.
Quick Facts about SUNET C
- Total price: ~300m SEK
- Backbone length: ~7000 kilometres
- Backbone capacity: 100 Gbps per wavelength
- Fibre types: G.652 and G.657
- Attenuation: max 0.25 dB/km, should be 40
- Contractual period: 11/01/2016 – 10/31/2031
- Delivery, huts and power: 01/31/2016
- Delivery, backbone: 03/31/2016
- Delivery, access network: 08/31/2016
- Delivery, additional transmission: 09/30/2016
- OptoSunet shutdown: 2016-autumn
- SUNET C in operation: 11/01/2016
- Acceptable downtime/month: 90 min
Mostly in Swedish. Please use Google Translate.
The Tele2 press release: http://om.tele2.se/tele2-bygger-sunets-nya-nationella-datafibernat-for-sveriges-hogskolor-och-universitet/
This is how Computer Sweden reported the purchase: http://computersweden.idg.se/2.2683/1.622696/universiteten-far-nytt-supernat-nar-sunet-storsatsar-pa-svartfiber
This is how Computer Sweden reported about network redundancy: http://computersweden.idg.se/2.2683/1.626414/sunet-bygger-monsternatet-som-ska-vara-framtidssakrat
About fibre of type G.652: http://www.itu.int/rec/T-REC-G.652-200911-I
About fibre of type G.657, bending-loss insensitive: https://www.itu.int/rec/T-REC-G.657-201210-I/en