Nobel WP2/A2.3/D31 Impact of differentiated CoR on network resources

1. Nobel WP2/A2.3/D31Impact of differentiated CoR on network resources Nobel Munich meeting June 13-15, 2005 Marco Quagliotti

2. Statement of the problem • In the past the transport networks normally offered two kinds of resilience options to the circuits to be carried: unprotected and protected (the last normally adopt 1+1 protection, with requires dedicated back-up resources) • The introduction of more sophisticated management, control and signalling capabilities in transport networks has made possible • restoration as a resilience mechanism. Restoration allows a saving in back-up resources (back-up resources are shared between failure disjoint connections) • offering circuits as extra-traffic (i. e. traffic that is not guaranteed in case of failure, even if the failure does not directly affect the circuit) and this allows to exploit back-up resources, which are unused in nominal (i. e. absence of failures) conditions • The objective of the study is to evaluate the impact of different QoR strategies on the amount of network resources (number of required channels -> indications for Capex)

3. Main hypotheses • A quantitative study has been performed by means of network dimensioning with different CoR scenarios and comparing them from the point of view of resource requirements • The reference network chosen is the Italian Long Distance Transport Network with 38 nodes and 79 links • The demand is composed of 1246 bidirectional circuits. It reflects more or less the actual traffic requirement in the Italian Network with a correction in order to obtain a more homogeneous/less polarized traffic distribution. All the circuits are supposed to be homogeneous in bandwidth (STM-16, 2.5 Gbit/s) • Four levels of class of resilience (CoR) are supposed to be required by the customers (and offered by the operator in the more sophisticated option) and a given partitioning of the whole traffic between the four classes has been assumed • No hypothesis on opacity/transparency or on other technological aspects has been made (the network is a generic transport network)

4. Classes of Resilience • Traffic is divided into Low and High quality traffic; High and Low refer to quality of Resilience requirement • Low quality is divided into 2 subclasses E - Extra traffic: connection is established on shared back-up resources for the working path and it does not have a back-up path -> connections that can be unavailable frequently and for long periods N – Unprotected: connection is established on dedicated resources for the working path and it does not have a back-up path -> connections that can tolerate a certain unavailability • High quality is divided into 2 other subclasses R - Restored: connections established on dedicated resources for the working path and on shared back-up resources for the secondary path -> connections that require a low unavailability but tolerate a non-instantaneous restoration time P – Protected: connection established on dedicated resources for both the working and back-up path -> connections that require a low unavailability and a quick restoration time

5. Scheme of network resources occupation for different CoR DEDICATED N-Unprotected working P- Protected DEDICATED back-up working R- Restored SHARED back-up E- Extra-Traffic Network resources

6. CoR Traffic distribution LOW Quality of resilience HIGH Class of Resilience assignment to the circuits has been done randomly with the constraint to assure a given CoR partitioning (E 10%, N 20%, R 50%, P 20% )

7. CoR scenario alternatives • Only one class is available; all the traffic is carried as protected in order to match the higher quality requirement (rather unrealistic) • Two classes are available: unprotected for low quality and protected for high quality traffic (used to be typical in the past for transport network) • Three classes are available: the restored class is added to the previous two in order to differentiate the high quality traffic into two subclasses (typical in present-day networks which support restoration) • Four classes I: extra-traffic is added to collect the lower quality requirement traffic within low quality traffic. Extra-traffic relies on restoration back-up resources • Four classes II: the same as before but extra-traffic relies on both restoration back-up and protection back-up resources (it is expected to be more efficient than the previous one)

8. CoR scenarios alternatives

9. Results (figures) Additional resources to back-up pool to cope with extra traffic

10. Results (diagram) - 35 % -17 % - 21% -15 % - 6 % ~ = Number of channels/(Net resources) Number of classes of Resilences / CoR scenarios

11. Conclusions • A comparison between less convenient (but unrealistic) CoR scenario (P only) and the most convenient one (N+P+E+R) shows a difference in terms of resources of about 35 % in total and of 53 % of back-up resources • A more significant comparison between the two-CoR scenario (N+P) and four-CoR (N+P+E+R) shows a saving of 21 % in total and 32 % of back-up resources • extra-traffic (here 10% of the whole traffic ) is almost completely absorbed by back-up resources. No significant differences between case 4 and 5 emerged • The more the traffic requirement matches with the applied QoR, the more the efficiency in network resources to be allocated, so the performed study strongly supports the differentiation of CoR in offering transport services • The study is performed on a single realistic network scenario and adopting a single CoR traffic partitioning but the feeling is that the indications can be considered suitable in a wider range of situations