1. 19.09.2005WP2 TEA1 Feasibility of Bandwidth on-Demand Case study approach, models and issues Stuttgart meeting 19-20 th of September 2005 WP2.3/2.5 Håkon Lønsethagen (Astrid Solem, Borgar Olsen) Telenor R&D

2. 19.09.2005WP2 TEA2 Outline Motivation and objectives of BoD case study Significant OPEX and CAPEX elements OPEX model CAPEX and OPEX related to Network and Service Management Traffic sources and assumptions Resource efficiency gains Tariffs and income Further work

3. 19.09.2005WP2 TEA3 Motivation and objectives of BoD case study Scope –Network operator deploys BoD and/or “dynamic” (L1) TE capabilities –Given the context of a L1-L2 network offering L1 and CO-PS services, typically L1 and Ethernet services –Services offered to Business customers, Service providers, and internal services Motivation –More efficient resource usage –OPEX and CAPEX savings –Increased Income Objectives –To explore and analyse issues related to feasibility of BoD Find significant OPEX and CAPEX changes compared with traditional leased lines and with ASTN based leased lines Find potential network resource efficiency gains Find potential increased demand for bandwidth Find potential new revenue and income

4. 19.09.2005WP2 TEA4 Significant OPEX and CAPEX elements that should be studied Making some rough estimates on which are significant OPEX and CAPEX elements OPEX –Service activation, deactivation and bandwidth upgrade/change –Equipment and SW licences –Charging and billing –Maintenance of equipment and components –Network management (e.g. SLA assurance) –Sales and marketing, Customer acquisition –Customer care CAPEX –Control plane –IT/OSS integration and new SW and licences Customer provisioning Service management, activation and assurance Control Plane management Charging and billing Customer care

5. 19.09.2005WP2 TEA5 Initial focus – Service management and bandwidth change –The most significant OPEX element –Have used Siemens/IMEC OPEX model cost elements for traditional and ASTN service provisioning –ASTN OPEX cost assessment vs. BoD OPEX cost assessment –Using connection length and bandwidth categories (granularities) OPEX model

6. 19.09.2005WP2 TEA6 SMo CEUCnR OPEX model – Service management SOCECSASCh SCe SOService Offer CEC Customer Edge (CE) Configuration/Provisioning SAService Activation SChService (Bandwidth/SLA) Change SCeService Cease CEUCE Upgrade CnRContract Renegotiation SMoService Move

7. 19.09.2005WP2 TEA7 No separate CAPEX-model developed The existing German reference network is taken as a basis in order to derive parameters for the model, that is: –The total traffic to be handled –Number of hops used to model ”lengths” –Number of connections The registration of types and number of leased line services in Telenor equivalent to the categories: –n*10M –n*100M –n*1G –Lambda Use of existing work Service categories Hamburg Berlin Hannover Bremen Norden Essen Dortmund Köln Düsseldorf Frankfurt Nürnberg Mannheim Karlsruhe Stuttgart Ulm München Leipzig

8. 19.09.2005WP2 TEA8 Services The registration of types and number of leased line services in Telenor equivalent to the categories: n*10M n*100M n*1G Lambda Is used to find the % of bandwidth consumed by each traffic category And thus the number of connection of each traffic category in a network handling the traffic according to the German network

9. 19.09.2005WP2 TEA9 Traffic sources and assumptions Hamburg Berlin Hannover Bremen Norden Essen Dortmund Köln Düsseldorf Frankfurt Nürnberg Mannheim Karlsruhe Stuttgart Ulm München Leipzig Total traffic - 3035 Gbit/s 2Mbit/sX% …. 2,5Gbit/sY% LambdaZ% Service categories n*10Mbs n*100Mbs n*1Gbs n*Lambda % of traffic for different service categories Scaled to network of German network size

10. 19.09.2005WP2 TEA10 Number of service changes % of traffic for different traffic categories Number of connections for different traffic categories Growth rate BoD Case: Percentage of connections in each traffic category and provisioning option Provisioning options: Freely allocatable Customer decided Leased lines Yearly number of connections for different traffic categories Scaled to German network Traditional case: Holding time BoD Case: % of service changes for each traffic category and provisioning option Traditional case: % of service changes

11. 19.09.2005WP2 TEA11 Total cost for service change Number of service changes for each traffic category For each traffic category the percentage of short, medium and long connections are the same –Percentage of connections with different lengths is derived from German network Short: 1-2 hops Medium: 3-4 hops Long: >4 hops The unit cost for traditional and BoD service change (depending on granularity/service category and length)

12. 19.09.2005WP2 TEA12 Cost per service change It is assumed that cost per service change will depend on length and traffic category Input: Cost per service change Lambda traditional and ASTN (from Siemens/IMEC work) Service change unit cost - Traditional case –A granularity multiplier is used to derive cost for the other traffic categories, resulting in cost for medium length connections for all granularities –Cost for short and long lengths for Traditional case is derived using the Length multipliers traditional Service change unit cost - BOD case –Derived from Unit cost traditional and BoD/Traditional cost reduction factors for BOD medium length –Unit cost for short and long is derived from BOD unit cost medium length through use of BoD length multipliers Method TO BE FURTHER ASSESSED –BoD is likely to require a greater cost reduction, compared with the Siemens/IMEC study, in order to be attractive

13. 19.09.2005WP2 TEA13 Unit cost for service change Initial “guesses”:

14. 19.09.2005WP2 TEA14 Further work – Input needed Consolidated NOBEL cost model –Further work needed (?) Control Plane cost model –OK? –Running/Support licence costs? CAPEX and OPEX related to IT/OSS –Initial cost, initial licence cost, system integration/modification cost, running licence cost (+ other IT/OSS OPEX cost) –Service Activation New and/or system integration/modification –Service Assurance New and/or system integration/modification –Charging New mediation components and system integration/modification –Billing System integration/modification Development of a NOBEL IT/OSS cost model? –Who can provide what input?

15. 19.09.2005WP2 TEA15 BOD Resource efficiency (CAPX) gains Motivation for BOD is to utilize temporal traffic variations between given end-points: Typically in the time range of minute – month (limited by packet switching in the lower limit and leased lines in the upper limit) Competing technologies are Leased lines (based on SDH or ) Packet switching (e.g. MPLS-based VPNs) Possible result: BoD is feasible if BoD for L1/L2 and BoD for L3/L2 is introduced into the market

16. 19.09.2005WP2 TEA16 BoD performance gain Gain potential BoD efficiency somewhere between leased lines and packet switching Granularity in time and capacity determine where: –frequent (relative to traffic variations) changes in small steps ~ packet switching –infrequent changes in large steps ~ leased lines Gain will vary significantly depending on traffic sources –10% - 20% - 40% ??? Interesting types of traffic variations Periodic (e.g. daily) variations, e.g. different busy hours between –Residential and business data traffic –Between ISP traffic and mobile traffic Large timescale stationary stochastic variations Existing results??

17. 19.09.2005WP2 TEA17 Pricing model – Assumptions and approach To develop a Pricing model for BoD services Expected increase in demand due to price reduction Approach: –Based on existing number of connections for traditional case: –What is the income we have to match (per hour?) –With the number of connections for BOD case – what amount of cost reductions can be achieved? Make a scaling matrix for tariffs with length and capacities relative to a given tariff (reference: n*10M and short set to 1) Multiply with the number of LL traditional contracts Do the same for BoD but distribute the tariffs according to holding times or 1/ frequency of changes, possibly taking into consideration categories (Freely allocatable, Customer decided and LL types) Normalise BoD tariffs to match LL traditional contracts to find the typical tariffs for the BoD services

18. 19.09.2005WP2 TEA18 Income – assumptions and approach (Example) Scaling factor for services: Revenue of LL 90 days as reference Nr of LL as reference: Revenue of BoD To be done!!

19. 19.09.2005WP2 TEA19 Further work – Open issues Tariff models Rough estimation of potential resource efficiency gains Demand changes – as a result of BoD Estimation of service change unit costs Rough estimation of IT/OSS related costs Analysis and summary of the initial rough estimations and results