ObjectOriented Design and Implementation of Fault Management Function

Object-Oriented Design and Implementation of Fault Management Function for MPLS network Sung-Jin Lim , Ryung-Min Kim, and Young-Tak Kim Advanced Networking Technology Lab. Dept. of Information & Communication Engineering Graduate School Yeungnam University, Korea Email: genie@yumail. ac. kr, degger@nownuri. net, ytkim@yu. ac. kr http: //antl. yu. ac. kr

Introduction • Traffic engineering has been emphasized to offer end-to-end Qo. S-guaranteed multimedia services in Next Generation Internet. • Fault restoration in traditional IP network vs. MPLS network • The primary goals of fast restoration by fault management function. – Qo. S guaranteed differentiated path protection. – Guaranteed bandwidth of backup LSP at fault occurrence. • We propose a fault management with fast rerouting restoration scheme in MPLS network. – Design and implement differentiated path protection and link. preemption priority among LSPs. • Key technologies – Traffic Engineering, Differentiated Service, Restoration, Protection, and Object-Oriented Design. APNOM 2003 (2) Yeungnam Univ. ANTLab.

Related works • Fault restoration model – Rerouting vs. protection switching • Rerouting • Protection switching : – Pre-established backup path SRLG-disjoint with working LSP – 1: 1, 1: N, M: N, 1+1 path protection switching – Local repair vs. path protection (A) Link protection (B) Node protection (C) Path protection Figure 1. Protection switching model APNOM 2003 (3) Yeungnam Univ. ANTLab.

Related works (cont. ) • Fault Management Architecture of Next Generation Internet – Fault Management Components in TINA – Fault Management MOs in TMN • Fault Management Activities – – – Alarm surveillance Testing Fault localization Fault correction Trouble administration • MPLS Fault Management System – RATES, – Cisco MPLS Tunnel Builder, – Sheer Networks’ Broadband Operating Supervisor(BOS) APNOM 2003 (4) Yeungnam Univ. ANTLab.

Restoration Schemes • Differentiated Path Protection Option – Example scenario of applying protection path options according to MPLS service class. Table 1. Differentiated Path Protection scheme MPLS Service Class Bandwidth Reservation Setup Priority Preemption Priority Application Platinum 100%, 1+1 Highest High Priority VPN Gold 100%, 1: 1 Higher Qo. S-guaranteed VPN Silver 100%, M: N Normal Premium service Bronze 100%, 1: N Lower Controlled traffic Best effort 0 Lowest Best Effort • Preemption Priority based restoration of LSPs – Guaranteed bandwidth of backup LSP for the protected working LSP. APNOM 2003 (5) Yeungnam Univ. ANTLab.

Design and Implementation of Fault Management System • Fault restoration procedure with Managed Objects (MOs) Figure 2. Fault Restoration on MOs APNOM 2003 (6) Yeungnam Univ. ANTLab.

Design and Implementation of Fault Management System (cont. ) • Fault Restoration Function – SNMP trap handler implementation – Alarm Manager GUI Figure 3. Event Log GUI Figure 4. Alarm Log GUI APNOM 2003 (7) Yeungnam Univ. ANTLab.

Design and Implementation of Fault Management System (cont. ) • Design and Implementation of NMS Core Figure 5. MO Classes of MPLS LSP Fault Management APNOM 2003 Figure 6. MO Classes of link/node Fault Management (8) Yeungnam Univ. ANTLab.

Performance Analysis of Fast Restoration • Test Network Configuration – Three core routers LER_E, LER_ F, and LER_G – Four CE router CE_A, CE_ B, CE_ C, and CE_ D – Two link types • POS (Packet Over SONET ) with 155 Mbps – Solid line • Serial with 2 Mbps – Dashed line Figure 7. Test network configuration APNOM 2003 (9) Yeungnam Univ. ANTLab.

End-to-end Performance Comparisons of Differentiated Protection Options(1) • End-to-end Performance Comparisons of Differentiated Protection Options Figure 8. Fast-reroute by NMS Figure 10. Standby mode by Cisco APNOM 2003 Figure 9. M: N fast-reroute by NMS Figure 11. Path-Option mode by Cisco (10) Figure 12. Link-Protection by Cisco Yeungnam Univ. ANTLab.

End-to-end Performance Comparison by Differentiated Protection Options • Differentiated Restoration performance by precedence Table 2. Priority and bandwidth between LSPs Working LSP (Tunnel 100) Low-Priority LSP (Tunnel 120) Same priority–based restoration Bandwidth sub-pool 1500 (backup LSP : Tunnel 110) Bandwidth sub-pool 1500 Different priority-based restoration Bandwidth sub-pool 1500 (backup LSP: Tunnel 110) Bandwidth sub-pool 1500 Figure 13. Same priority APNOM 2003 Note (After failure occurrence) - does not satisfy required bandwidth - does not create backup TE-LSP - Working traffic: Transfer to dynamic LSP - satisfy required bandwidth - Preemption link with high Priority - create backup TE-LSP Figure 14. Different priority (11) Yeungnam Univ. ANTLab.

Conclusion • We proposed an object-oriented design and implementation scheme for – 1: 1, 1: N, M: N fast-reroute by NMS – Standby mode, Path-option and link/node protection scheme in Cisco MPLS Routers. • The Proposed Fault Management Scheme for MPLS Network Provides – Reliability that guarantee the required bandwidth of backup LSP after fault restoration – Differentiated protection path option – Object-Oriented MO design and implementation of network nodes and links for better expansibility with equipments form various vendors. APNOM 2003 (12) Yeungnam Univ. ANTLab.