RIPE NCC Quarterly Report Q2, 1996 Ridley ____________________________________________________ Reseaux IP Europeens Network Coordination Centre Quarterly Report Second Quarter 1996 Paul Ridley with contributions from Daniel Karrenberg Mirjam Kuhne Carol Orange Document: ripe-148 ____________________________________________________ ripe-148.txt Page 1 RIPE NCC Quarterly Report Q2, 1996 Ridley ____________________________________________________ 1. Introduction RIPE is a collaborative organisation which consists of European Internet service providers. It has been functioning since 1989. Today nearly 3.3 million computers in Europe and surrounding areas are reach- able via networks coordinated by RIPE. RIPE aims to provide the necessary administrative and technical coordination to allow the operation of a seamless pan-European IP network. The RIPE NCC The RIPE Network Coordination Centre started opera- tions in April 1992. It supports RIPE activities that cannot be effectively performed by volunteers from the participating organisations. Key services performed by the RIPE NCC include: o Maintenance of the RIPE Network Management Database o Regional Internet Registry o Domain name system (DNS) coordination o Repository for network operations software o Maintenance of the RIPE document store o Interactive Information Services The TERENA association provides the formal framework for the NCC. Funding for the operation of the NCC is provided by the Internet service providers in Europe and surrounding areas. General information about the NCC, its funding, organisation and activities can be found in the fol- lowing documents: Doc ID Title -------------------------------------------------------------- ripe-125 RIPE NCC Activity Plan ripe-133 RIPE NCC Activities & Expenditure 1996 ripe-134 RIPE NCC Revenue & Charging 1996 ripe-132 RIPE NCC Contributors Committee, Minutes Sept 1995 ____________________________________________________ ripe-148.txt Page 2 RIPE NCC Quarterly Report Q2, 1996 Ridley ____________________________________________________ Further information can be obtained by sending mail to or contact us at: RIPE Network Coordination Centre Kruislaan 409 NL-1098 SJ Amsterdam The Netherlands Phone: +31 20 592 5065 Fax: +31 20 592 5090 E-Mail: http://www.ripe.net The Q2-1996 Quarterly In this report, we aim to inform the RIPE community in general and the RIPE NCC Contributors Committee in particular of our activities during the second quarter of 1996 and our plans for the next one. Due to a resource shortage and a change over of positions in the third quarter this quarterly report is somewhat delayed in arrival and somewhat reduced in content. The Q3-1996 report will however be pro- duced on time and will be more complete. We welcome any comments and suggestions to further improve the usefulness of these reports. ____________________________________________________ ripe-148.txt Page 3 RIPE NCC Quarterly Report Q2, 1996 Ridley ____________________________________________________ 2. Management Summary The European Internet continues to grow at a phenom- enal rate. During Q2-1996 57 new local Internet registries started operations. This growing number of local registries proportionately increases our workload. During Q2-1996 alone, we manually handled 9,770 messages and automatically handled 26,913 mes- sages that were sent to the RIPE NCC role mailboxes. Registration services continued to operate normally, albeit with a constant backlog due to factors including increased demand and the training of new staff. The two newly hired hostmasters started work in Q2-1996 and have immediately settled in and made valuable contributions. We continued gathering hostcounts, providing local registry courses and information services on the Internet. Significant activities during the quarter were among others; the publication of ripe-136 the "European Internet Registry: IP Address Space Assignment Poli- cies and Procedures", which supersedes ripe-104; the continued development of the 2.0 release of the database software, planned to be released in Q3-1996; the implementation of the hierarchical authorisation mechanism for the inetnum and domain database objects; the launch of the ripe ncc web site; the start of the project looking into usage based charging models. The financial situation is very healthy with more than 80% of this year's planned expenditure already received in revenue. During Q3-1996 significant activities will include; pursuance of the stated staffing plan; giving more training courses for local Internet registries; pre- senting various usage based charging models to the contributors; publication of the 1997 activities and expenditure plan; and publication on the ftp site of individual allocation statistics. Strategically we observe that the Internet continues to change too quickly for stringent 12 month plan- ning cycles. We will carry on in our efforts to adapt more quickly to these changes. Considering the rapidly changing environment it sup- ports and operates in, the RIPE NCC is healthy and functioning well. ____________________________________________________ ripe-148.txt Page 4 RIPE NCC Quarterly Report Q2, 1996 Ridley ____________________________________________________ 3. Core services In this section, we describe the activities in NCC core services which were carried out during Q2-1996. 3.1. Role Mailbox Traffic The NCC handles requests and queries almost exclu- sively by electronic mail. A number of role mail- boxes are used for this purpose. These mailboxes are regularly serviced by the appropriate NCC per- sonnel or automated procedures. Personal mailboxes are rarely used for request handling. While serves as a catch-all for all sorts of queries and requests, the other mailboxes are intended for specific types of requests. A full description of the following mailboxes can be found in the Q1-1996 report (ripe-135): In addition to those listed above we have a number of role mailboxes dealing with meeting and course registrations. In the table below you find counts of the number of messages being received by the various role mail- boxes during Q2-1996. ____________________________________________________ ripe-148.txt Page 5 RIPE NCC Quarterly Report Q2, 1996 Ridley ____________________________________________________ +--------------------------------------+ |Role Mailbox Activity Q2-1996 | +--------------------------------------+ |Mailbox Messages | +--------------------------------------+ |ncc 3338 | |billing 977 | |hostmaster 3233 | |inaddr 1339 | |ripe-dbm 883 | |auto-inaddr 2209 | |auto-dbm 24704 | +--------------------------------------+ A total of 9770 messages were sent to the manual mailboxes in Q2-1996, and the messages sent to auto- matic mailboxes totaled 26913. As may be expected, of the manual mailboxes, by far the most messages are sent to and . On average, there are 150 messages sent to the man- ual role mailboxes each working day, most of which originate from NCC customers. Meanwhile, an average of 414 messages are sent to the automatic mailboxes each day. 3.2. Registration Services During Q2-1996 seven staff members were involved in Registration Services, namely: Paula Caslav, John Crain, Mirjam Kuehne, Nick Reid, Els Willems, Lee Wilmot. Paula Caslav and Lee Wilmot started in April to replace Soodabeh Eshgi and Hatice Kuey who left their posts at the RIPE NCC during Q1-1996. More details on the status of affairs in Registra- tion Services in mid-April, 1996 can be found in: ftp://ftp.ripe.net/ripe/presentations/ripe- m24-mir-RS-REPORT.ps.gz 3.2.1. Hostcount One of the services provided by the RIPE NCC is the regular hostcount report which tracks the growth of ____________________________________________________ ripe-148.txt Page 6 RIPE NCC Quarterly Report Q2, 1996 Ridley ____________________________________________________ the Internet in Europe and surrounding areas. Here we present a brief summary of our recent findings. In the following graph, we demonstrate the exponen- tial growth in the number of hosts connected to the Internet in Europe over the last five years. The number of hosts connected to the Internet in Europe continued to grow during Q2-1996, as illus- trated in the graph below. Some of the anomalies in the hostcount statistics are caused by connectivity problems. Others are caused by neglect of DNS zone maintenance. In both cases, the data is usually corrected in later months with the growth following its general exponential behaviour. The hostcount is based on the number of hosts counted in 53 DNS Top Level Domains (TLD's). The complete set of TLD's contributing to the statistics shown here are listed in Appendix A. Historically, the majority of the hostcount statis- tics have been gathered remotely by the RIPE NCC. Due to network overhead, and the anomalies which occur due to connectivity problems, an effort is underway to have hostcounts performed locally per TLD and to compile the results at the NCC. At the end of Q2-1996, local counts are performed in 23 TLDs, 2 more than at the end of Q1-1996. The RIPE NCC still performs 30 hostcounts remotely, so this effort will continue for the foreseeable future. 3.2.2. Number of Local IRs In Q1-1996 the growth in new registries increased by more than 30 percent. This trend continued through- out Q2-1996. We still saw almost 20 new registries each month. In the table below, we show the number of registries of varying types and sizes for each quarter in 1995 and for Q1-1996 and Q2-1996, along with the pro- jected numbers for the remaining quarters in 1996. The projection we made in the first quarter of 1996 showed a stabilisation of the the growth rate to about 18 new registries a month. These projections were it seems too conservative. The same growth pat- tern can be seen in Q2-1996 and is expected to ____________________________________________________ ripe-148.txt Page 7 RIPE NCC Quarterly Report Q2, 1996 Ridley ____________________________________________________ continue for the remaining quarters of 1996. This would then lead to almost 600 registries at the close of 1996, nearly twice as many as at the close of 1995. +--------------------------------------------------------------------+ | Number of Local IRs | +------------+-----------------------------------------+-------------+ |Type | OBSERVED | PROJECTED | | | Q1 Q2 Q3 Q4 Q1 Q2 | Q3 Q4 | | | 1995 1995 1995 1995 1996 1996 | 1996 1996 | +------------+-----------------------------------------+-------------+ |Large | 17 19 25 28 32 34 | 37 41 | |Medium | 31 35 36 40 49 53 | 60 65 | |Small | 84 119 159 196 246 308 | 376 451 | |Enterprise | 15 17 15 16 19 26 | 31 37 | |Last Resort | 32 32 30 28 22 13 | 6 0 | +------------+-----------------------------------------+-------------+ |TOTAL | 179 222 265 308 368 434 | 510 594 | +------------+-----------------------------------------+-------------+ 3.2.3. Registration Services Usage Statistics Hostmaster Workload To give an indication of the workload in the host- master arena throughout the quarter, one can view the following graph in which we show the number of incoming messages to per day during Q2-1996. In the next graph we show the number of new requests opened per day in the hostmaster mailbox. Note that this is lower than the number of messages as many messages may be sent back and forth regarding a sin- gle request. ____________________________________________________ ripe-148.txt Page 8 RIPE NCC Quarterly Report Q2, 1996 Ridley ____________________________________________________ Finally, we show the level of the hostmaster wait queue during Q2-1996. This is a daily count of the of requests which are opened but not yet assigned to a specific hostmaster for processing. The wait queue is a good measure of the responsive- ness of the service. It is clear that current staffing in Registration Services is just sufficient to handle the load, but insufficient to provide pre- dictable service in the face of unexpected shortages and increasing demand. In the course of Q2-1996, we have continued to track the Registration Services workload, and started col- lecting information about the nature and frequency of different types of requests being submitted. This workload data was collated and a short report was presented by an MBA student as part of his study in a usage based charging model. Thereby we were able to quantify how much work we perform in different areas, to identify the most effective task areas for further automation, and draw staffing conclusions. This short report can be found at Appendix B. 3.2.4. Policies and Procedure Documentation An updated version of ripe-104 the "European Inter- net Registry: IP Address Space Assignment Proce- dures" was published as ripe-136 "European Internet Registry: Policies and Procedures". Also the "Euro- pean IP Address Space Request Form" (ripe-137) and the accompanying "Supporting Notes for the European IP Address Space Request Form" (ripe-138) have been updated and obsolete ripe-128 and ripe-129. 3.2.5. Last Resort Registries It has been decided by the Contributors Committee to only provide service to contributing Local Reg- istries. This means that most Last Resort registries will have to be closed if they don't decide to con- tinue their service and contribute to the RIPE NCC. At the end of Q2-1996 2 LIR's have comitted to con- tinue their service and another 10 have been closed. The remaining 15 Last Resort registries do not receive service anymore and an effort will be made to formally close them as soon as possible. This involves updating the records on all sites and often the return of address space and reverse delgation. ____________________________________________________ ripe-148.txt Page 9 RIPE NCC Quarterly Report Q2, 1996 Ridley ____________________________________________________ 3.3. Services for New Registries As detailed in Section 3.2.2, 57 new local IRs started operations during Q2-1996. To assure these new local IRs understand and operate according to the policies agreed to by the RIPE community, we provide them with extra support in the startup phase. Our support for new registries falls into two pri- mary categories, namely individual help in opening the registry, and the local IR training courses. 3.3.1. Help with Local IR Startup The process of setting up new registries is handled by the RIPE NCC billing department. It involves some administrative steps including the signing of a for- mal agreement. In this phase, we try to assure each new registry gains access to basic information on IP address space allocation and assignment procedures. Organisations that consider becoming a local IR usu- ally send questions concerning the usefulness of setting up a registry and the consequences it has for them. Questions are received on the kind of ser- vice the RIPE NCC offers, formal agreements, IP address space allocation policies and Internet rout- ing issues. Whenever these organisations come to the RIPE NCC, the billing department tries to address their ques- tions and to introduce those asking to the standard documentation that new registries should be aware of. 3.3.2. Local IR Training Program During Q2-1996, we delivered 3 Training Courses for Local IRs, namely: Q2-1996 Trainings ------------------------------ May 10, 1996 Budapest May 20, 1996 Amsterdam June 7, 1996 Stockholm We allow a maximum of 15 persons to attend each course, and all three of the above were filled. To ____________________________________________________ ripe-148.txt Page 10 RIPE NCC Quarterly Report Q2, 1996 Ridley ____________________________________________________ measure the effectiveness of our training course, we ask all persons attending to fill in an evaluation form. The results show our courses to be perceived well and considered useful. Because of the rapid increase in the number of local IRs, our courses fill up very quickly upon being announced. We thus recognise the need to increase the frequency of these courses. 3.4. The RIPE Network Management Database One of the key services provided by the RIPE NCC is the maintenance of the RIPE database software. In this section we report on progress made in this area during Q2-1996, as well as a number of interesting database statistics gathered on June 30, 1996. 3.4.1. Recent Progress During Q2-1996, development continued in the prepa- ration of the 2.0 release of the database software, planned for Q3-1996. The most important work which took place during the second quarter was the implementation of the hierar- chical authorisation mechanism for the inetnum and domain database objects. ____________________________________________________ ripe-148.txt Page 11 RIPE NCC Quarterly Report Q2, 1996 Ridley ____________________________________________________ 3.4.2. Usage statistics In the following table we show the number of entries for the different objects maintained in the RIPE database at the end of Q2-1996. RIPE Database Object Number ---------------------------------------- as-macro 85 aut-num 802 community 7 domain 41761 dom-prefix 6 inetnum 52047 inet-rtr 99 limerick 10 mntner 646 person 76347 route 12948 ---------------------------------------- TOTAL 184758 During Q2-1996, 2,721,150 queries were received by the NCC whois server. 3.5. RIPE Meetings RIPE meetings are held three times a year. Tradi- tionally two are held in Amsterdam and one is hosted by an organisation elsewhere in Europe. The meeting consists of working group sessions that focus on specific topics, and a plenary session with general presentations. RIPE meetings are open to everyone interested in them. 3.5.1. RIPE-24 The 24rd RIPE meeting was held April 22nd-24th 1996 in Berlin. There were 130 attendees. The minutes of the meeting are available from ftp://ftp.ripe.net/ripe/minutes/ripe-m-24.txt ftp://ftp.ripe.net/ripe/minutes/ripe-m-24.ps A number of RIPE NCC staff members made presenta- tions at the meeting. The transparencies used in ____________________________________________________ ripe-148.txt Page 12 RIPE NCC Quarterly Report Q2, 1996 Ridley ____________________________________________________ the presentations can be found in: ftp://ftp.ripe.net/ripe/presentations/ripe- m24-david-DB-REPORT.ps.gz ftp://ftp.ripe.net/ripe/presentations/ripe- m24-mir-RS-REPORT.ps.gz 3.5.2. RIPE-25 Preparations are underway for the 25th RIPE meeting to be held September 23rd-25th 1996 in Amsterdam. Up-to-date information can be found at the URL ftp://ftp.ripe.net/ripe/Next-Meeting/ ____________________________________________________ ripe-148.txt Page 13 RIPE NCC Quarterly Report Q2, 1996 Ridley ____________________________________________________ 4. RIPE NCC Staffing As planned in Q1-1996 two new hostmasters, Paula Caslav and Lee Wilmot, started work in this quarter. Since these hiring were direct replacements for the two hostmasters who left in Q1-1996 they do not effect operations at the RIPE NCC. A detailed organ- isational structure was published in the Q1-1996 report (ripe-135) and remains unchanged. 4.1. NCC Staff Planning The staffing plans outlined in the Q1-1996 report were pursued in quarter 2 but without a great deal of success in finding suitable candidates. In quar- ter 3 different avenues will be pursued and the tempo will be upped in a determined effort to achieve the laid down plans. ____________________________________________________ ripe-148.txt Page 14 RIPE NCC Quarterly Report Q2, 1996 Ridley ____________________________________________________ 5. Financial Summary This financial summary will follow the format pre- sented in the quarter 1 report. At present we are restricted by the data that can be gathered from our administrative system but once this is revised more detail and depth of reporting will be possible and will be carried out. It should also be noted that if comparisons are made between this quarterly report and the TERENA half yearly report then apparent dis- crepancies will be found. In truth there are no dis- crepancies but only differences in reporting proce- dures. This quarterly report follows the RIPE NCC tradition of reporting with the assumption that the period end is the end of the financial year, whereas TERENA reports assume that the period end is the end of the respective quarter. This difference will man- ifest itself by TERENA only reporting the income received for those services already given whereas this report includes income received for services yet to come in the financial year. 5.1. Quarter Summary: Q2-1996 At the start of Q2-1996, there were a total of 368 local Internet registries in the RIPE NCC service area. Now there are 421, and given that another four last resort registries have closed shop during this time, we've had a total of 57 new local IRs starting operations during Q2-1996. The expenditure budget for 1996 approved by the RIPE NCC contributors committee in September 1995 is 916 kECU. At the end of Q2-1996 a total of kECU 385 had been expended, equated to 42% of the budget. It is to be expected that expenditure in the first half of the year will be less than the second half since the planned expansion will cost proportionately more money. It should be noted though that the computer and travel expenditure was slightly higher than forecast and these individual budgets may need to be amended in Q3-1996. At the end of Q2-1996, had commitments for 904 kECU of revenue, which is 99% of the expenditure budget. At this point kECU 761 has been received. Therefore, 83% of the currently planned yearly expenditure has been received in the second quarter. This all results in the NCC having a very healthy liquidity situation. A project was also started in Q2-1996 to look at the ____________________________________________________ ripe-148.txt Page 15 RIPE NCC Quarterly Report Q2, 1996 Ridley ____________________________________________________ possibilities of a usage based charging system. This project is being performed by two MBA students. The results of this report will be ready by the begin- ning of Q3-1996 and will form the based of the 1997 revenue and charging document. If the growth in the number of IRs continues at its present pace, our income for services performed in 1996 will exceed 1500 kECU and thus be 60% above currently planned expenditure. Of course this increase in demand will also require additional resources. Should this trend continue in Q3-1996 we will compile a revised budget to cope with it. ____________________________________________________ ripe-148.txt Page 16 RIPE NCC Quarterly Report Q2, 1996 Ridley ____________________________________________________ 6. Q3-1996 6.1. Efforts and Expectations Local IR Training Without exception all local IR courses are immedi- ately filled up. To try and better cater for this high demand we are increasing the number of sched- uled courses in Q3-1996 to six. These are listed in the table below. Q3-1996 Trainings --------------------------------- July 8, 1996 Milan July 19, 1996 Geneva August 9, 1996 Amsterdam August 30, 1996 Prague September 16, 1996 Vienna September 28, 1996 Amsterdam If possible, additional trainings in Q3-1996 will be scheduled. As always, we are happy if a local IR offers to host a course. However due to the high demand, we have decided to make an effort to hold courses frequently whether or not we have a host. To make it easy for people to attend these courses, most will be held at or near major airports in Europe. The courses for Q4-1996 will be announced to the local IR mailing list as usual. RIPE Database Developments We will continue a concerted effort to complete ver- sion 2.0 of the database software, and will release the software during Q3-1996. After the release future RIPE database developments will be assessed and prioritised. Documentation Both the "1997 Activities and Expenditure Plan" and the "1997 Revenue and Charging Model" will be ____________________________________________________ ripe-148.txt Page 17 RIPE NCC Quarterly Report Q2, 1996 Ridley ____________________________________________________ completed in Q3-1996. These documents will then be presented to the contributors, at the contributors meeting on 11 September 1996, for their approval. The "Revenue and Charging Model" will contain sev- eral possibilities from which the contributors will have to choose the most suitable. Allocation Statistics In response to a request made in Q1-1996 we will, in Q3-1996, publish a list of individual allocation statistics. This will be published on the ripe ftp site. Staff Hiring During Q3-1996 we hope, in accordance with the staffing plan, to hire four new staff members, with the following job titles: o 2 Hostmasters (to keep registration services running smoothly) o Network Engineer (to help in the effort to automate services) o Manager - Systems and Software (to lead automation and systems activities) ____________________________________________________ ripe-148.txt Page 18 RIPE NCC Quarterly Report Q2, 1996 Ridley ____________________________________________________ Appendix A: Hostcount TLD's The set of DNS Top Level Domains contributing to the hostcount statistics described in Section 3.1.1 are: al: Albania am: Armenia at: Austria az: Azerbaijan be: Belgium bg: Bulgaria by: Belarus ch: Switzerlan cy: Cyprus cz: Czech Republic de: Germany dk: Denmark dz: Algeria ee: Estonia eg: Egypt es: Spain fi: Finland fo: Faroe Islands fr: France gb: United Kingdom ge: Georgia gr: Greece hr: Kroatia hu: Hungary ie: Ireland il: Isreal is: Iceland it: Italy li: Liechtenstein lt: Lithuania lu: Luxemburg lv: Latvia ma: Morocco md: Republic of Moldova mk: The Former Yugoslav Republic Of Macedonia mt: Malta nl: The Netherlands no: Norway pl: Poland pt: Portugal ro: Romania ru: Russian Federation se: Sweden si: Slovenia sk: Slovakia (Slovak Republic) sm: San Marino su: The Former Soviet Union tn: Tunesia tr: Turkey ua: Ukraine uk: United Kingdom va: Holy See (Vatican City State) yu: Yugoslavia ____________________________________________________ ripe-148.txt Page 19 RIPE NCC Quarterly Report Q2, 1996 Ridley ____________________________________________________ Appendix B: Registration Services Project During the Q2-1996 an MBA student started conducting a study into a usage based pricing system for RIPE NCC. As a part of this study the activities and work practices of the Registration Services department were analysed. The results of this analysis give a valuable insight into the percentage of time spent on various activities and show the relationship between Registration Services' present workload and capacity. Based upon these new insights we have been able to draw three points for future operations. Whenever messages come into they are categorised according to which of the eleven possible activities they relate. Therefore the number of times that a certain activity is done per day can be readily seen. A part of the study was to determine the average time that it took to com- plete each individual activity through the use of activity flowcharting. By combining this average time data with the activity frequency data it was possible to ascertain how much time should have been spent, per day, dealing with the messages that came into Hostmaster. It was noteworthy that Assignment Window Exceeded requests are by a sizeable margin the most time consuming activity performed. IN-ADDR also occupies a large amount of the working hours. Obviously no week is exactly the same as the previ- ous one, thus in order to gain a more in-depth look at the weekly workload of Registration Services, information was gathered concerning the number of hours per week that were totally dedicated to these eleven activities. From this information it was concluded that at the beginning of the second quarter the average number of hours per week that needed to be spent on the eleven activities was 80, but at the end of the quarter it has risen sharply to an average of 110 hours. With present staffing levels for the depart- ment of one manager and 4.6 fte hostmasters it is possible to dedicate an average of 85 hours per week to eleven core activities. An average fte hostmaster will dedicate 18.5 hours per week to these activi- ties. From figure (capacity chart) it can be readily seen that with the rare odd exception the department has been receiving a workload that is in excess of its present 85 hour capacity. This growing gap between workload and capacity has led to the inevitable increase in wait time that the requests are experiencing. The present shortfall between ____________________________________________________ ripe-148.txt Page 20 RIPE NCC Quarterly Report Q2, 1996 Ridley ____________________________________________________ workload and capacity is 25 hours per week which is equivalent to 1.4 fte employees. From the information provided by this study into the Registration Services department we have arrived at three ways of improving service and cutting costs. Firstly the activity flowcharting that was carried out has added more insight into the operational flow and has further highlighted areas that would benefit from a higher degree of automation. Therefore based upon the extra insight we plan to refine our automation program. Secondly since the Assignment Window Exceeded activity occu- pies such a large amount of work time we are looking at ways to minimise this. Thirdly since it has been shown that there is a definite discrepancy between the present workload and capacity of the registra- tion department we need to hire two extra fte host- masters. This would allow the increasing wait times to be attacked, thus improving service. As growth of RIPE NCC continues, and the workload proportionally increases, the situation of the Registration Ser- vices department with regard to staffing will have to be kept under review. ____________________________________________________ ripe-148.txt Page 21