Abstract- Due to routing failure there is failure of packet delivery which is common on Internet and routing protocol cannot always react fast to recover from them. To provide solutions on this problem fast reroute solution have been proposed to guarantee reroute path availability and avoid high packet loss after network failure. It is difficult to provide solution to protect internet for both intra and inter domain routing due to their individual computational and storage complexity. In particular, enhanced protection cycle (e ‘cycle) is proposed for both intra and inter domain routing protocol that used to construct rerouting path and provide link and node protection. In this paper, we consider intra domain routing and try to optimize virtual cycle required for e-cycle. E-cycle adopt p-cycle (virtual cycle) which take long path under failure .We try to optimize virtual cycle required e-cycle by using Integer Linear Programming Algorithm. We evaluate our solution by simulation.
Keywords: Routing, Routing protection, virtual cycle.
1. Introduction
Today many people are using internet. They are participating in real time network application such as online gaming, online transaction, entertainment and e-commerce application. Failures occur at various protocol layers in the network due to different reasons. At the physical layer, due to a fiber cut or a failure of optical equipment there is to loss of physical connectivity. Hardware failures (e.g. line card failures), router processor overloads, software errors, protocol implementation and misconfiguration errors may also be the cause loss of connectivity between routers. When network components (such as routers, line cards, or optical fibers) are shared by multiple IP links, their failures affect all the links. Failures may be occur due to unplanned scheduled network maintenance. To connect different IP Networks there is requirement of internet routing which play critical role in ensuring packet delivery throughout the internet. From previous study it is clear that current routing systems are ineffective to protect against failure [1]. There is problem of slow convergence of distance vector (DV) routing algorithms. It requires that each node maintain the distance from itself to each possible destination and the vector, or neighbor, to use to reach that destination. Router transmits its new distance vector to each of its neighbors when connectivity information changes and also allow allowing each node to recalculate its routing table .After topological changes DV routing can take a long time to converge after a topological change[2] The link-state routing protocols such as OSPF is commonly deployed in today’s networks which react to link failure, disseminate link-state changes, and then recomputed their routing tables using the updated topology information. In high speed networks even short recovery time can cause huge packet loss. In failure insensitive routing FIR, when a link fails, only nodes adjacent to it locally reroute packets to the affected destinations and all other nodes simply forward packets according to their precomputed interface-specific forwarding tables without being explicitly aware of the failure. Once the failed link comes up again, forwarding resumes over the recovered link [3].
Boarder Gateway Protocol (BGP) session reset and transient hardware failure. It suffers from both frequent routing changes and slow convergence. While there have been previous efforts in inferring the location of routing changes the root cause of the routing changes are extremely hard to infer since many different causes, such as physical layer failures, link layer failures, configuration changes, congestion and router software bugs, can lead to the same routing updates.BGP routing updates often provide little support in investigating these causes[4] To recover from such failure current routing protocol unable to react quickly .There are many researcher provide solutions in order to effectively address routing failure. They focus on fast routing convergence [5]. But none of these solutions has been deployed in operational network due to their complexity. Furthermore, the additional computation complexity introduced by having multiple interacting routings also makes it unlikely that standard, or even similar solutions as used for a single routing, can be readily applied. Computational complexity arises from two main sources, one of which is already present in standard IP routing (single routing). Specifically, upon detecting failures, IP routing adjusts its packet forwarding decisions, i.e., recomputed shortest paths based on the configured link weights; to better redistribute traffic around the failure [6].There is requirement of extra route withdrawal message in Ghost Flushing for convergence in failover event. Fast routing convergence is ineffective for handling routing black holes and loops. It is important to provide routing protection by using backup routing paths in case of routing failure but this have some design limitations. IP FRR [7] mainly focuses on intra domain routing. It is a potential technique to improve IP resilience in intra-domain routing, which can switch traffic to backup routes quickly. In general, IPFRR approaches can be implemented by different backup path selection algorithms. To reduce the computation overhead, several improved tunnel-based IPFRR solutions are proposed such as reduce the number of shortest path tree (SPT) computations with the Not -via approach in IPFRR [8] BGP-FRR proposed to support fast reroute in inter domain routing from previous studies it is clear that there was protection for single type of routing either intra or inter domain routing.
In this paper, we proposed enhanced protection for routing in IP network. We try to optimize virtual cycle required for e -cycle. E-cycle adopts p-cycle which is large under failure. In p-cycles, the network is covered by a set of logical rings. Each ring protects all links it contains, ‘swaddling’ links, i.e., links between non-neighboring nodes within the ring. It is not specified how the rings are to be arranged on a topology, and there are many possible layouts. A layout can be created manually or by a centralized algorithm. A single cycle incorporating all nodes in the network is one possible p-cycles layout .p-Cycles redirects the network traffic so it follows the cycle layout. In the case of large cycles, this is expected to cause long protection paths [9] .
So we try to optimize that virtual path required for e-cycle using ILP algorithm .Which will reduced packet loss, increase throughput, reduced delay and also provide full coverage .this will achieve using simulation
2. Problem in Existing Fast Reroute Solutions
There are many solutions for fast reroute. That forward packet along alternate path under network failure, then provide protection and improve performance. Some solutions are ineffective due to their computational complexity. Enhanced e-cycle is proposed to construct rerouting path and provide link and node protection for both intra and inter domain routing. Enhanced cycle adopt p-cycle called virtual cycle which provides practical and lightweight solution routing protection and it is first design for failure recovery in SONET and WDM. To provide fast reroute for node and link failure recovery using p-cycle requires minimal candidate virtual cycles.[2]
However, p-cycle has some drawbacks .The length of rerouting path in original p-cycle solution significantly enlarge failure, as the packet go through the whole remainder of cycle and forward on based on normal cycle. It follows long routing path under failure so need to minimize virtual cycle required for e-cycle which will increase the throughput and load of the network will increase that increases performance which show better result.
3. RELATED WORK
Previous auto-discovery protection solutions were more complex. Enhanced protection cycle (e-cycle) provides t pre-configured routing paths to realize fast rerouting efficiently. To construct rerouting path e-cycle uses virtual cycle like p-cycle Similar to p-cycle and also uses different identifiers (e cycle IDs) to identify rerouting paths. Thus provides protection for all nodes and links. E-cycle different from previous technology the difference is that, since every router has routes to destinations, e-cycle does not detour packets along an entire virtual cycle as in p-cycle, but tries to find an forwarded along normal routes. E-cycle uses two components, protection initiators (PIs) and protection terminators (PTs). Protection initiators (PIs) are routers who detect failures and then activate protection paths to forward packets, and protection terminators (PTs) are routers who terminate protection paths and continue normal packet forwarding.
After construction of e-cycle we have to select a PT for every PI in the cycle1 and e-cycle ID based forwarding is only applied along the partial cycle between PI and PT. When PI detects a failure, it starts to forward affect packets along the e-cycle towards PT. The new packet header contains an e-cycle ID field which specify the unique identifier of the e-cycle used for fast forwarding, and a hop count field which specifies the hop count between PI and PT. The hop count indicates the lifetime of the packet in the e-cycle, and will decrease by one when that packet is forwarded by a router. If the hop count equals to zero, the packet will be removed from the e-cycle by PT and the original packet will be forwarded along a normal route to its destination.[7]To enhance protection we tried to implement Integer Linear programming (ILP algorithm) to e-cycle that will improve the performanance of the system. This will discuss in the next section.
4. PROPOSED SYSTEM
Here we are considering an autonomous system.
Fig: 1 Autonomous System
.
Fig: 2 Autonomous system with link breakage
Fig 3: Protection cycle p-cycle under failure
In e-cycle, virtual cycle construction is an important procedure for deployment in operational networks. However, p-cycle has some drawbacks. The length of a rerouting path in the original p-cycle solution is significantly enlarge under failures. We try to optimize reroute path or virtual cycle required for e cycle which is enlarging under failure for intra domain system. This will achieve using Integer Linear Programming (ILP) algorithm.
Steps of ILP algorithm are follows.
[1] Go through each possible path.
[2]Find out link quality (given by NS2) and distance from source (found using Euclidean Distance formula).
Euclidean Distance = sqrt ((x2-x1) ^2 + (y2-y1) ^2)
[3]Combine the link quality and distance using formula.
Alpha and beta have default standard values of 0.5
score = (alpha*link quality) + (beta/distance)
Alpha and Beta are constants between 0 to 1
[4]For data transfer that node will be selected which having highest score.
References
[1] A. Markopoulou, G. Iannaccone, S. Bhattacharyya, C. Chuah, and C. Diot, ‘Characterization of failures in an IP backbone,’ in Proc. 2004 IEEE INFOCOM, pp. 2307’2317.
[2] C. Labovitz, A. Ahuja, A. Bose, and F. Jahanian, ‘Delayed Internet routing convergence,’ IEEE/ACM Trans. Networking, vol. 9, no. 3, pp.293’306, 2001.
[3] S. Nelakuditi, S. Lee, Y. Yu, Z. Zhang, and C. Chuah, ‘Fast local rerouting for handling transient link failures,’ IEEE/ACM Trans. Networking,vol. 15, no. 2, pp. 359’372, 2007.
[4] L. Wang, M. Saranu, J. Cottlieb, and D. Pei, ‘Understanding BGPsession failures in a large ISP,’ in Proc. 2007 IEEE INFOCOM.
[5] M. Shand and S. Bryant, ‘IP fast reroute framework’, RFC5714, Jan.2010.
[6] K. W. Kwong, R. Guerin, A. Shaikh, and S. Tao, ‘Balancing performance, robustness and flexibility in routing systems,’ IEEE Trans.Network and Service Management, vol. 7, no. 3, pp. 186’199, 2010.
[7]Qi Li, Mingwei Xu, Jianping Wu, Xingang Shi, Dah Ming Chiu and Patrick P.C. Lee, ‘Achieving Unified Protection for IP Routing’ ,IEEE pp. 978-1-4244-7116-4,2010.
[8]Y. Yang, M. Xu, and Q. Li, ‘A light-weight IP fast reroute algorithm with tunneling,’ in Proc. 2010 IEEE ICC.
[9] T. Cicic, A. Kvalbein, A. F. Hansen, and S. Gjessing, ‘Resilient routing layers and p-cycles: tradeoffs in network fault tolerance,’ in Proc. 2005 HPSR, pp. 278’282.