RIPE NCC Contributors Committee 1996 Annual Meeting ____________________________________________________ RIPE NCC Contributors Committee 1996 Annual Meeting Minutes Daniel Karrenberg Roderik Muit Document: ripe-145 The meeting was called by Steve Druck, TERENA vice president for services, and held on September 11th 1995 at the Dorint Hotel, Badhoevedorp, Nether- lands. These minutes were prepared by Roderik Muit and Daniel Karrenberg of the RIPE NCC. Participants Bernward Averwald:de.roka Lajos Balint:hu.hungarnet Per Bilse:eu.eunet Jaroslav Bobovsky:sk.sanet Daniel David:de.ipf Steve Druck:TERENA (Chairman) Klaus Ellegaard:dk.azlan John Evans:fr.netsat Jan Gruntorad:cz.cesnet Hans Petter Holen:no.sn Daniel Karrenberg:RIPE NCC Alexandre Krivine:SKYWORLD:akrivine@sky.fr Mirjam Kuehne:RIPE NCC Wolfgang Mair:de.profinet Eric Merkoske-Bree:fr.netsat Dave Morton:de.ecrc Roderik Muit:RIPE NCC (minutes) Kees Neggers:nl.surfnet Mike Norris:ie.heanet Daniel Nussbaum:ch.tic Carol Orange:RIPE NCC ____________________________________________________ ripe-145.txt Page 1 RIPE NCC Contributors Committee 1996 Annual Meeting ____________________________________________________ Rick Payne:uk.netcom Marco Pomponi:it.asianbyte Marc Roger:be.belnet Nick Shield:uk.janet Daniel Sjoberg:se.telia Henk Steenman:eu.att Andrea Stefankovicova:sk.pubnet Bernhard Stockman:se.telia Karel Vietsch:TERENA Wim Vink:eu.eunet Bob Wardrop:fr.netsat Tony Weatherill:uk.mercury Ton Windgassen:eu.ibm Wilfried Woeber:at.aconet Apologies The following people had sent their apologies prior to the meeting: Paul Ridley: RIPE NCC Gregoriy Dmitriev:ru.fastnet Giuliana Tamorri:it.garr Konstantin Vinogradov:ru.fastnet Richard Oxby:fr.netsat Bill Unsworth:uk.u-net Roger Doire:fr.cybercable Pontus Ekman:se.pi 1. Welcome & Preliminaries Steve Druck welcomed the participants and was elected to chair the meeting. Wolfgang Mair suggested to add an extra agenda point about a new action of the German authoroties con- cerning Internet regulation. Daniel Karrenberg pointed out that the contributors were not the right group to take actions on this: it would be interest- ing to hear something about the developments but actions or decisions should be taken by RIPE. It was decided not to discuss this matter. Steve Druck suggested that the recently published registry procedures document (ripe-140) could be discussed by the meeting. Unfortunately this docu- ment was not ready in time for this discussion to be ____________________________________________________ ripe-145.txt Page 2 RIPE NCC Contributors Committee 1996 Annual Meeting ____________________________________________________ put on the agenda, hence formal endorsement by the contributors should be done via e-mail. Wilfried Woeber noted that the changes from RIPE-136 to RIPE-140 were mainly editorial, and RIPE-136 had already been adopted by the Local IRs. He concurred that endorsement by the contributors was useful and should occur via the mailing list. No-one expressed the need for further discussion. Therefore the dis- cussion was not added to the agenda. During discussions later that day, issues were brought up that were moved towards the end of the meeting, and added as separate agenda points 6 and 7. Agenda The final agenda for the meeting: 1 Welcome, Selection of Chair for Meeting Steve Druck, TERENA VP for Services 2 NCC Activity Report Q1-Q3/96 Daniel Karrenberg, Mirjam Kuehne, Carol Orange For information and comments. 3 NCC Financial Report Q1-Q3/96 Daniel Karrenberg For information and comments. 4 NCC Activities and Expenditure 1997 Daniel Karrenberg Proposal sent out August 30th. For discussion and decision. 5 RIPE NCC Revenue & Charging 1997 Carol Orange Proposal sent out September 3rd For discussion and decision. 6 Level of Reserves Added during the meeting. 7 New Legal Structure for the RIPE NCC Added during the meeting. 8 Closing ____________________________________________________ ripe-145.txt Page 3 RIPE NCC Contributors Committee 1996 Annual Meeting ____________________________________________________ 2. NCC Activity Report, Q1-Q3/1996 Documents "RIPE NCC Annual Report 1995" ftp://ftp.ripe.net/ripe/docs/ripe-139.ps ftp://ftp.ripe.net/ripe/docs/ripe-139.txt "RIPE NCC Quarterly Report Q1-1996" ftp://ftp.ripe.net/ripe/docs/ripe-135.ps ftp://ftp.ripe.net/ripe/docs/ripe-135.txt The RIPE NCC reported on activities since the last meeting. The growth in the number of local reg- istries has again been significantly higher than estimated. The NCC has been able to keep up with essential services but coordination and new activi- ties have suffered. As suggested by the contributors the NCC's structure has been changed to include a business manager. In addition there is now an engineering manager in charge of all engineering activities. So the man- agement team now consists of Daniel Karrenberg:NCC Mirjam Kuehne:Registration Carol Orange:Engineering Paul Ridley:Business Please refer to the documents for details about spe- cific activities. Discussion Bernard Stockman asked how many registries a host- master can manage. Mirjam Kuehne answers that the NCC measures this not in terms of registries but in terms of specific activities the hostmasters per- form, like assignments, allocations etc.. Wilfried Woeber asked if the RIPE NCC received feed- back from the registries about the plan to assign hostmasters specifically to registries. Mirjam answered that the NCC has received quite positive reactions about this plan. ____________________________________________________ ripe-145.txt Page 4 RIPE NCC Contributors Committee 1996 Annual Meeting ____________________________________________________ Bernard Stockman asked what the RIPE NCC considers an acceptable delay in response time for requests sent to hostmaster@ripe.net. Daniel Karrenberg replied that this is not for the NCC to decide but for the local registries. From feedback he got from the registries, it came out that two to three days was considered still acceptable and the response time should not grow above that. Hans Petter Holen noted that the crucial thing is not as much the response time as it is knowing what the response time will be. He would like to see a way in which Local IRs can know this in advance. Mirjam Kuehne replied that the RIPE NCC has thought about this fact and will implement a scheme for this as part of the improvements to be made to the request tracking system. 3. NCC Financial Report Q1-Q3/96 Documents "RIPE NCC Annual Report 1995" ftp://ftp.ripe.net/ripe/docs/ripe-139.ps ftp://ftp.ripe.net/ripe/docs/ripe-139.txt Daniel Karrenberg gave an overview over the current financial status of the RIPE NCC including a revised budget for 1996. Daniel concluded that the financial position of the NCC is currently very healthy. Daniel added that future financial reports will be made in the form of more established financial reporting formats such as profit/loss accounts and balance sheets. Discussion Wim Vink asked how it could happen that the RIPE NCC is suddenly being exposed to 1995 company tax. Daniel Karrenberg and Steve Druck stated that not budgeting for this was obviously an oversight of the TERENA Executive Committee, the TERENA Secretariat and the RIPE NCC. The TEC, the Secretariat and the NCC acknowledge this mistake. Tony Weatherill noted that it would be a good thing for people to look into the RIPE NCC financial plans and make suggestions. He called on the Contributors as a group to feel more responsible. ____________________________________________________ ripe-145.txt Page 5 RIPE NCC Contributors Committee 1996 Annual Meeting ____________________________________________________ Mike Norris added that indeed the budgets had been presented to the contributors and noone noted the omission. He also added that he considered the right actions had been taken by adding a business manager to the NCC team. Note: A discussion evolved about the budgeted level of reserves for 1997 and possible ways for the RIPE NCC to avoid taxes in the future, which was added as a separate agenda point but still carried on throughout the day. All of this dis- cussion is summarised at point 6 in an attempt to make the minutes more consistent.) Hans Petter Holen stated that the Contributors Com- mittee should endorse the revised budget for 1996 in this meeting. Steve Druck noted that the budget had been published on the mailinglist and nobody objected, so in fact it had been approved. Hans Pet- ter Holen insisted that the Contributors should actively approve the budget, because that is part of their role. This was also moved to the end of the meeting. Hans Petter wanted to know what the role of the con- tributor's committee was. If they would not make decisions along these lines, then what were they here for? What exactly is the role of the co-co? That discussion was moved to the end of the day. 4. NCC Activities and Expenditure 1997 Documents "RIPE NCC Activities & Expenditure 1997" ripe-act ftp:...drafts/ripe-act.... Daniel Karrenberg began by stressing that the main challenge to the RIPE NCC is to manage its growth. He compared the growth estimates made at the previ- ous meeting with the actual figures which again were higher than the projections. He then gave a short overview over the document including the current growth projections, as well as the activities and resources needed. ____________________________________________________ ripe-145.txt Page 6 RIPE NCC Contributors Committee 1996 Annual Meeting ____________________________________________________ Discussion Steve Druck stated that the RIPE NCC would be approached as a candidate for operating a European CERT coordination centre. He asked the audience whether they thought the RIPE NCC was a suitable place to operate this. Wilfried Woeber supported this idea fully. Kees Neg- gers noted that this would change the scope of the RIPE NCC, since it is not a monopoly activity. Steve Druck answered that it is also a coordinating activ- ity. Also, it would be financed independently. To Bernard Stockman's question whether the RIPE NCC was interested, Daniel Karrenberg replied that he did not know yet what the proposed activity was about. He added that the NCC should probably be involved because the customers for a CERT coordina- tion centre are mainly the ISPs which the NCC already serves and has built relationships with. The experience the NCC has with large scale coordi- nation efforts could also be useful to this activ- ity. Wim Vink asked who would approach the NCC as a can- didate and Steve Druck said that TERENA would. In the first instance, a European CERT coordination centre would be a TERENA activity but it was expected that it would eventually stand on its own. Tony Weatherill supported the idea, but noted that the best thing to do was first investigate the scope of a CERT coordination centre and to be wary of the possibility that the NCC might loose its current focus. Steve Druck confirmed that the proposal was at this moment for a pilot organisation of one to three people. There was consensus a CERT activity could be exe- cuted at the NCC if the NCC can accommodate it and it funded separately from other NCC activities. The contributors requested to be kept informed about developments. A second question from Steve Druck concerned whether the contributors thought that the RIPE NCC could play a role as representing the European Internet Industry. ____________________________________________________ ripe-145.txt Page 7 RIPE NCC Contributors Committee 1996 Annual Meeting ____________________________________________________ Mike Norris proposed that activity N.4, Industry Representation, was more a coordination activity. Daniel Karrenberg stated he had no idea which way the discussion about this issue inside RIPE would go and there was no concrete plan for this activity yet. The general opinion expressed by the meeting was that the RIPE NCC could do this only on matters that are inside the scope of RIPE NCC's current activi- ties, such as address space registration and provi- sion of Internet statistics. Care should be taken not to become defocussed. The RIPE NCC should not represent the European Internet industry as a lobby- ing organisation because that industry usually has no collective opinion on many matters. Taking stands here could endanger the RIPE NCC's neutral position. The idea to set up an advisory committee inside RIPE, which advises the RIPE NCC on matters of industry representation, was dismissed. Steve Druck finalised the agenda point by asking everyone whether they approved the Activities & Expenditure document. This was done and the RIPE NCC was asked to amend the plan according to the discus- sion and publish it. 5. RIPE NCC Revenue & Charging 1997 Documents: "Alternative Models for RIPE NCC Revenue & Charging 1997" ftp://ftp.ripe.net/ripe/docs/ripe-143.ps ftp://ftp.ripe.net/ripe/docs/ripe-143.txt Carol Orange presented the document including the proposed revenue and the three alternative charging models. Discussion Steve Druck asked Daniel Karrenberg to summarise the discussions on the contributors mailing list. Daniel said that he had had no opportunity to ____________________________________________________ ripe-145.txt Page 8 RIPE NCC Contributors Committee 1996 Annual Meeting ____________________________________________________ perform exact counts. His estimate of responses both to the contributors list and to him privately was that about 80% of the people preferred model 1, with model 3 as second choice, and about 20% pre- ferred model 3, with model 1 as second choice. There was very little support for model 2 and some rejected it explicitly. Other concerns expressed were the administrative overhead that quarterly billing would bring. Daniel noted that these concerns could easily be addressed by appropriate invoicing procedures for any of the models proposed. Some small registries complained about the increased charges for their category. After short discussion there was consensus not to pursue model 2 any further and to see whether con- sensus could be established to use either model 1 or model 3. What was then discussed next came down to various properties of model 3: Wilfried Woeber asked why the charge for a new allo- cation of a /16 in the first quarter of 1997 is so much higher than for an allocation in the fourth quarter. Carol noted that the algorithm is explained in the document, but if the contributors felt this should be changed, then the difference could easily be flattened. It was noted that model 3 encourages conservation of address space, which is a good thing. It was also noted that it is somewhat at odds with address space aggregation. The model should be adjusted in a way that aggregation is not hurt too much. Mike Norris expressed concern that there will be a charge linked specifically to an allocation of address space, while RIPE just agreed that IP addresses have no intrinsic value. Carol Orange responded that the charge was designed in every pos- sible way to be not for the addresses per se but for the registration services associated with it. This was evident from the fact that the charge decreases over time as the expected level of services needed does. After further intensive discussion it became clear that consensus to use either model 1 or model 3 as proposed could not be achieved. Many contributors found model 3 to be too complex in that computing their own fee would require a calcu- lator, spreadsheet or program to determine. Some ____________________________________________________ ripe-145.txt Page 9 RIPE NCC Contributors Committee 1996 Annual Meeting ____________________________________________________ also felt the administrative side is too complex and that all parameters need to be agreed upon. A sim- pler model 3 was argued for. The suggestion of a flat fee per allocation was however turned down, because the RIPE NCC has much less work on older allocations. An alternative suggestion was made to use model 1 with 3 fixed categories, but to determine a minimum size category according to the address space allo- cated to each registry. This would determine the minimum charge for a given registry but still allow the registry to choose a larger category if they wished. This was supported by most of those pre- sent. Daniel Karrenberg stated that he does not want the RIPE NCC to make subjective decisions on the size of registries. He wanted to have an algorithm to decide the category and requested some input on the number of categories as well as the parameters to be used to decide the minimum registry size. He also suggested to use this algorithm as an advice and leave the choice up to the registries, as not all registries can be judged in the same way. The over- all reaction from the meeting was that the choice should not be left up to the registries and the min- imum size category should be binding. The discus- sion on the number of categories came up several times more during the meeting and it was concluded that the number should remain three. Kees Neggers stated that the contributors should explicitly decide that the important parameter in the algorithm is load on the RIPE NCC, not address space allocated. Per Bilse and Wilfried Woeber said that the most expensive activities out of model 2 should go into the algorithm. Daniel said he would surely consider this and noted that the most impor- tant item was the training courses. The NCC would continue to consider obtaining revenue directly from the courses while at the same time encouraging their use. Per Bilse suggested to make training courses manda- tory for new registries before they can start inter- action with the RIPE NCC. After some discussion Daniel Karrenberg stated that this was not practi- cally feasible for 1997 as the NCC could not guaran- tee to make sufficient training capacity available. He added that he was not currently in favour of a what could be called "Certified Registry Operator". ____________________________________________________ ripe-145.txt Page 10 RIPE NCC Contributors Committee 1996 Annual Meeting ____________________________________________________ Kees Neggers' suggestion that the algorithm be decided upon in the next meeting met opposition because it was felt that the relevant research had already been done. Some also stated that given the expected growth rate it is beneficial to make rea- sonable changes earlier rather than later. Kees also suggested to only apply this algorithm to new registries and to leave currently existing reg- istries in their current categories the next year. This was turned down after Daniel said he expected a lot of controversy especially from new registries and it would generate a lot of discussion. After further thorough discussion the RIPE NCC com- mitted to have an algorithm for determining sizes ready by October 1st and suggested would circulate the proposal on the mailinglist by then. Wim Vink suggested that the algorithm should just be imple- mented right away. The final decision was that the RIPE NCC would design and implement the 1997 charging scheme with- out further endorsement from the contributors. The scheme would be based on the principles described in ripe-143 and model 1 in particular. The NCC would design an algorithm to determine a registry's size and therefore minimum charging category. The algo- rithm would be a simplified version of the model 3 algorithm with three size categories. The fees must be set so that any cross subsidies are kept at an absolute minimum. Via the mailing list, the con- tributers will be notified of the charging scheme, the registry sizes and fees by October 1st 1996. 6. Level of Reserves The discussion on the amount of budgeted reserves had come up several times. Some of those present felt that the amount of reserves was too high, espe- cially since a lot of taxes has to be paid over the surpluses needed to build them. Karel Vietsch explained that in its current legal setup there was no way the RIPE NCC could get out of paying Dutch company tax on its surpluses. There was extensive discussion whether or not the different categories of reserves were actually over- lapping, and whether the reserves could be covered by the registries paying in advance or whether pay- ing in advance was only an issue of cash flow and not of balance. ____________________________________________________ ripe-145.txt Page 11 RIPE NCC Contributors Committee 1996 Annual Meeting ____________________________________________________ The suggestion was made by several people that the RIPE NCC should not have to have a high level of reserves, as it could always fall back on the con- tributors in case of troubles. Karel Vietsch and Daniel Karrenberg were concerned that getting extra money from all contributing Local IRs would cause a lot of problems since a lot of Local IRs see them- selves more as customers than as members, and the current contract, while having words to support this, is not explicit enough. Karel expressed the concern that in case of trouble, the contributors might drop the RIPE NCC, form another company and leave TERENA with the debts. He noted that this is a reason to think about making a legal separation between TERENA and the RIPE NCC. Very thorough and intensive discussions about the level of reserves and the way to obtain them fol- lowed. There were many suggestions including to have reserves in the form of subordinate loans from the contributors to the NCC. After the discussions it was agreed that consensus about a long term level of reserves could not be established at this meet- ing. The problem should be attacked in a structural way in conjunction with the development of a new legal structure. Consensus about both issues should be established in the coming months. Provided that this was an interim solution it was agreed for 1997 to budget for a reserve level of one year of salary cost to be earned over two years. 7. New Legal Structure for the RIPE NCC There was consensus among those present that the legal structure of the RIPE NCC should be changed in order to separate it cleanly from TERENA and min- imise tax liabilities. Karel Vietsch stated that TERENA fully agreed with this and wished to take an active role in the process. Daniel Karrenberg was very happy about the general consensus on such an important issue. He stressed that developing a new structure should be done carefully to safeguard the stability and independence of the NCC while giving the contributors the necessary control. After some discussion it was agreed that a logical and realis- tic target date to have the new structure opera- tional was January 1st 1998. Steve Druck moved the discussion forward to the sub- ject of finding a committee that would investigate possibilities for a new organisational structure for the RIPE NCC. He volunteered Karel Vietsch and ____________________________________________________ ripe-145.txt Page 12 RIPE NCC Contributors Committee 1996 Annual Meeting ____________________________________________________ himself as members. Carol Orange suggested Paul Rid- ley and Daniel Karrenberg as advisory (non-voting) members. Mike Norris suggested Rob Blokzijl. Con- tributors were still looked for. Wim Vink volun- teered on the condition that the committee would be very small, with 3 people maximum. Daniel Karrenberg noted that with such a small group of people, the decision making process would have to be very clear if general consensus among all contributors was ever to be reached. There was agreement that a small group should work to establish the general possibilities and report back to the contributors. The members of the group suggested were Wim Vink, Karel Vietsch and Paul Rid- ley. 8. Closing Steve Druck concluded that the 1997 activities and budget had been agreed with amendments concerning the level of reserves, that the NCC was given a spe- cific mandate to determine the 1997 charging scheme and that a new legal structure for the NCC would be developed. He thanked all participants for a constructive meet- ing and encouraged them to take an active part in the mailing list discussions over the coming months. ____________________________________________________ ripe-145.txt Page 13