Towards Expeditious and Unswerving Routing to Corroborate Nascent Internet
The internet is now-a-days experiencing a stress due to some inherent problems with the main interdomain routing protocol, boarder gateway protocol (BGP), the amount of time it takes to converge, number of update message exchanged followed by a failu…
Authors: Shishir Kumar, Mahesh Kumar
JOURNAL OF COMPUTING, VOLUME 1, ISSUE 1, DECEMBER 2009, ISSN: 2151-9617 HTTPS://SITES.GOOGLE.COM/S ITE/JOURNALOFCOMPUTING/ 114 T owards Expeditious and Unswerving Routing to Corroborate Nascent Internet Dr Shishir Kumar , Mahesh kumar Abst ract —The internet is now-a-days experiencing a stress due to so me inherent problems with the main interdomain routing protocol, boarder gateway pro tocol (BGP), the a mount of ti me it takes to converge, number of u pdate message exchanged followed by a failure to stabilize, t he amount of time required to get a valid alternate path following the failure, the way si ze of routing table increasing, and security issues like integrity and privacy of routing t ables and routing updates exchanged among the routers, are of our primary concern. In our proposed res earch work we plan to address aforem entioned issues related to internet routing specially in boarder gateway protocol to enabl e BGP to offer expeditious unswe rving routing to corroborate nascent internet. We plan to make some changes in the desi gn of boarder gateway protocol and may introduce additi on of extra features in BGP to help suppor t above mentioned objective. Index T erms — Computer Networks, internet routing, BGP , internet gr owth, routing protocols, rout ing tables, routing updates, convergence time —————————— —————————— 1 I NTRODUCTION HE structure of the internet is a is a collection of networks, or Autonomous Systems (AS’ s), as shown in figure1 and 2, which are interconnecte d to form a connected domain [19]. Each AS uses an interior routin g system to maintain a coherent view of the topology within the AS, and uses an exterior routing system to maintain adjacency information with neighboring AS’ s and thereby create a view of the connectivity of the entire system. This network-wide connectivity is described in the routing table used by the BGP4 protocol. Each entry in the table refers to a distinct route. The attributes of the route are used to determine the best path from the local AS to the AS that is originating the route. Determining the ’best path’ in this case is determining which routing adver- tisement and associated next hop address is the most pre- ferred. The BGP routing system is not aware of fine r level of topology within the local AS or within any remote AS. From this perspective BGP can be seen as a connectivity maintenance protocol, and the BGP routing table, a de- scription of the current connec tivity of the Internet, using an AS as the basic element of computation. Figure 1: Autonomous Systems T ——————————————— — • Dr Shishir Kumar is Head of the Department of Computer Science and Engineering, Jaypee Institute of Engineeringv and Technology, Guna, Madhya Pradesh (India). • Mahesh Kumar is with the Department of Computer Science and Engineer- ing, Jaypee Institute of Engineeri ngv and Technology, Guna, Madhya Pradesh (India). © 2009 Journal of Computing http://sites.google.com/site/jo urnalofcomputing/ JOURNAL OF COMPUTING, VOLUME 1, ISSUE 1, DECEMBER 2009, ISSN : 2151-9617 HTTPS://SITES.GOOGLE.COM/S ITE/JOURNALOFCOMPUTING/ 115 Figure 2: Multi-tier inter-dom ain routing 2 RELA TED WORK: Different authors have tried different approaches to overcome issues in internet routing problem s, and each one has claimed to achiev ed better performa nce by either introducing new approaches or by making some different modifications to one factor or two of the existing protocol, but as we know that every factor is not completely unrelated to others, so by just ignoring the impact of change in one parameter of the protocol on others is not convincing.So It becomes necessary to compare all the results together and see that when one factor is redu ced then its impact on all whole internet. Here its an attempt to study the wholesomeness internet performance and improve it. 3 CONVERGENCE TIME A study of packet delivery performance during routing convergence have shown that network failures happen frequently, and that existing routing protocols converge slowly after a failure. During these routing co nvergence periods, some packets may already be enro ute to their destinations and new packets may be sent. These on the way packets can encounter routing loops, delays , and losses [1]. Sometimes BGP takes a substantial amount of time and messages to conv erge and stabilize following the failure of some node in the Internet. A very common technique Route Flap Damping was introduced in BGP protocol to minimize the impact of relatively unsta ble routes, and almost all router manufactu rers use this ap ‐ proach in the ir routers. Cisco and Juniper use in their routers to deliberately delay route calc ulat ions to increase stability. But flip side of RFD is the that sometime it can delay the network converge nce, through simulation re ‐ sults it has been observed that route flap damping can significantly exacerbate the convergenc e times of rela ‐ tively stable routes[3]. A new mechanism BGP ‐ RCN, that provides an upper bound of O(d) on routing convergence delay for BGP, where d is the network diameter as meas ‐ ured by the number of AS hops. BGP ‐ RCN lets each rout ‐ ing update message carry the information about the spe ‐ cific cause which triggered the update message. Once a node v receives the first update message triggered by a link failure, v can avoid using any paths that have been obsolete by the same failure [10]. 4 SCALABILITY Whenever we try to reduce the amount of time taken to converge, we try to keep most of the information about alternative paths to the destination in the router . But if we do not carefully look at the entries coming into the rout ‐ ing table after a failure or change in network, then routing table entries may increase to a point where it becomes difficult to ma nage it. Network operators and dev elopers have shown their concern about the routing table grow ‐ ing at alarming rate which has potential to affect entire Internet. The Int erne t continues along a path of seemingly inexorable growth, at a rate that has, at a mini mum, dou ‐ bled in size each y e ar . How big it needs to be to meet fu ‐ ture demands remains an area of somewhat va g u e specu ‐ lation. Of more direct interest is the question of whether the basic elements of the Internet can be extended to meet such levels of future demand, whatev er they may be. To rephrase this question, are there inherent limitations in the technology of the Internet—or its architecture of de ‐ ployment—that may impact the continued growth of the Internet to meet ever ‐ expanding levels of demand [19]? The current Internet interdomain routing system will not be able to scale properly to meet growing needs, re ‐ searchers have not ye t produced any routing architecture with satisfactory app roach to limit the growth of afore ‐ mentioned entries. We believe that we will be facing in near future a challenge to scale the size of the Internet so as the size of ro uti ng table entries. © 2009 Journal of Computing http://sites.google.com/site/jo urnalofcomputing/ JOURNAL OF COMPUTING, VOLUME 1, ISSUE 1, DECEMBER 2009, ISSN: 2151-9617 HTTPS://SITES.GOOGLE.COM/S ITE/JOURNALOFCOMPUTING/ 116 Figure 3:BGP Routing Ta b l e Grow th Pattern The last critic al point wa s reached when the Internet’ s routing system adopted strong address aggregation using CIDR to handle address scaling in the mid 1990’ s. While CIDR wa s an extremely effective tactic, most experts agree that the growth beha vior of the rout ing table in the last decade conf irms that the type of address manag e ‐ ment require d by CIDR will not suffice to meet future Internet routing needs [1]. . To d a y the growing Internet needs a reasonably good ag ‐ gregation like CIDR has done last time which can redu ce the size of routing table and change the wa y the routi ng table grow . But we its not so easy because aggregation beyond a limi t can not help us as we know that it may raise issues load balancing and other traffic engineering. The problem with the Internet’ s rout ing architecture is how the interdomain rout ing proto col algorithm and its BGP implementation scale with the size of the network. Po o r scaling of a rout ing algorithm expresses itself in terms of rapid rates of growth of the routing table si ze. Po o r scal ing of routing table sizes exacerbates conver ‐ gence problem [5]. Not only is the communication over ‐ head of BGP known to be exponential [1], but the BGP RT size also appears to grow exponentially [5]. 5 FAULT MANAGEMENT Even as recently as a decade ago, the failure of an Internet based system wou l d hav e been a relatively mi ‐ nor annoya nce. To d a y , however , such failures have an enormous cost, make news headlines, and, above all, hav e serious consequences on our society . The coming ye a r s will see the reach of the Internet extending wider than ever before, and together with this increasing reac h will come a need for robust operation far more stringent than in the past. These dual trends, one a “technology push” towar d pervasiv eness driven by the integration of com ‐ puting, wireless communication, and sensing technolo ‐ gies on small devices, and the other a demand pull driv en by the use of networks as a critical component of the wo rl d ’ s information sys tems, form the backdrop for my research[7]. Fault ‐ tolerance is the ability to operate correctly under faulty conditions. These conditions in a network setting can include the failure of network components such as physical link failure, node failure, and switch failures. Components within nodes can become faulty . Fault in ‐ stances can be permanent or transient, or a combinati on of these. To d a y we need fault tolerance in internet rout ‐ ing. 6 ROUTING POLICIES The configurations of routing protocols determine how packets traverse each of these levels of Internet topology . A routing protocol is responsible for exchanging informa ‐ tion about the state of the network and deciding which paths to use to reach ev ery destination. The output of the routing protocol is a forwarding table. The primary role of a routing protocol is to detect and avo id failed links, but it also allows operators to express preferences for dif ‐ ferent paths to shape how traffic flows across a topology . Routing between domains is determined by policy . Each autonomous system can, based on configured policy , in ‐ dependently select routing information from its neighbouring autonomous systems, and selectively propagate this information. These policies are not ex ‐ pressed in term s of hop ‐ distance to destin ations [11]. De ‐ pending on how these policies are constructed, then, the resulting policy ‐ based paths to destinations may inc ur more ro uter level hops than shortest ‐ router ‐ hop path routing. Some paths may be preferred because they are lightly ‐ used, cheaper , or more reliable. The preferences for different paths constit ute routing policy , which ISP operators express in the configuration of rout ing proto ‐ cols [8]. 7 ROBUSTNESS AND SECU RITY Wit ho u t introducing a fool proof security mechanis m to protect routers and their routing tables it does not seem possible to maintain a co rrect flow of information among desired sources and destinations, extending the networks, and servi ces without compromises. But we also unde r ‐ stand that securing Internet routing is a challenging task. We need a flexible and scalable protocol and most impor ‐ tantly , a deployment strategy , since the Int ern et consists today of hun d reds of thousands of rout ers and tens of thousands of independent networks [13]. We have seen from the pre vious experiences that a single malfun ction ‐ ing router can poison the rout ing tables of many oth e r routers on the network. There va ri o us issues like prefix hijackin g, unauthor ‐ ized advertisement of IP prefixes, use of illegitimate paths, spamming, wor ms , trojan horses, password ma ‐ nipulating, and phishing which we wish to include in our © 2009 Journal of Computing http://sites.google.com/site/jo urnalofcomputing/ JOURNAL OF COMPUTING, VOLUME 1, ISSUE 1, DECEMBER 2009, ISSN : 2151-9617 HTTPS://SITES.GOOGLE.COM/S ITE/JOURNALOFCOMPUTING/ 117 research work to ha ve reliable interdomain routing proto ‐ col. 8 PROPOSED METHODOL OGY In our ap proa ch, we plan to achieve expeditious and unswerving routing R based of the following concept: R=f(T ,S,F ,Se,P o) Where: T is the convergence time of boarder gateway protocol, S is the size of the rout ing table of boarder gateway protocol, F is fault tolerance factor of boa rder gateway protocol, Se is the security factor of boa rder gateway protocol, and Po is the routing policy factor on boarder gatewa y protocol. That is our proposed routin g approach is to optimize each above mentioned factor of boarder gateway protocol to get expeditious and unswerving routing decisions. To reduce the convergence time of boarder gatewa y protocol we plan to adopt a me thod to suggest some changes in design of boarder gateway protocol like treat ‐ ing the updates differently based on their nature and the ver y purpose. We plan to simulate the internet topology with many IS Ps connected together and then by injecting faults with the help of a separate node into the simulated environment and will observe the its time of convergence, and then we will modify SSLD, classi fy the updates, MRAI timer mechanisms to get reduced convergence time. We will run simulations large number of time to get relatively mo re accuracy if we get expected result the changes will be accepted otherwise some other changes will be done in the design. An approach to internet routing scala bility and flexibil ‐ ity is to sepa rate the identification and locator of a node in the network. In an architecture where one label identifies a node and a different label indicates its loca ‐ tion, topological changes will only change the locators which are assumed to follow topology and allow for ag ‐ gregation, and then we pl an to introduce an addition translation mechanis m, after labelling a tag to each ro uter belong to a particula r area, which can translate it to nor ‐ mal identification which can be easily identified by route r of different area. Our p roposed approach is to have a separate table of disjoint paths in a sepa rate database in boarder gateway protocol for each possi ble sourc e ‐ destination pair if pos ‐ sible without strictly sticking to the condition of selecting shortest path. Whenever a failure oc curs in that situation the priorit y wou ld be to get a new path from this database of disjoint paths table to forw ard traffic on alternate path. To provide security and robustness to boa rder gateway protocol we prop ose a strict met hod of content checking on each update receiv ed from neighbours, and if its con ‐ tent shows intuition of ma l ‐ intent then that update needs to be discarded and all the paths going through that router should be dropped from the database and a data ‐ base of black listed router should be edited. 9 CONCLUSION While studying the behaviour of boarder gateway pro ‐ tocol (BGP) we found that the convergence time, the number of update messages exchanged among routers, and the size of the rout ing tables increase at a rapid rate whenever network experience any change in the network connectivity/link failure. The entire network becomes over burd ened and congested while processing these re ‐ quests, we sa w that sometime even a sin gle rout e with ‐ drawal message can trigger hundreds of rout e withdraw ‐ als and new route advertisements and the whole network of rout ers become overloaded with this wor k of stabiliz ‐ ing that user traffic has to suffer during this period. In our proposed wor k we will limit the rate at which these withdrawal and advertisements emerge following a change in network connectivity so that the network can stabilize faster and will be available mo re to support user traffic. We also proposed met hods to make BGP robus t and secure by having a separate table of disjoint paths in a separate database for each possi ble source ‐ destination pair if possibl e without strictly sticking to the condition of selecting shortest path. Whenev er a failure occurs in that situation the priority wou l d be to get a new path from this database of disjoint paths table to forward traffic on alternate path. R EFERENCES [1] Crag Labovitz, Abha Ahuja, Abhijit Bose, Farnam Ja hanian “De ‐ layed Internet Routing Convergence ” in Proc. of ACM SIG ‐ COMM’00 , Aug. 2000. [2] Dan P ei, Lan Wa n g , Daniel Massey , et. al., “A Study of Pa c k e t Delivery P erfo rmance During Routing Convergence”, Proceed ‐ ings of the 2003 IEEE International Conference on Dependable Sys ‐ tems and Networks(DSN ʹ 03) , Pa g e ( s) : 183 ‐ 192, Ye a r :2003. [3] Zhuoqing Morley Mao, Rames h Govindan, George Varghese, and Randy H. Katz “Route Flap Damping Exacerbates Internet Routing Convergence” ., Proceedings of the 2002 SIGCOMM conference. Pages: 221 ‐ 233 Year of Publication: 2002 [4] Zhi ‐ Li Zhang Jaideep Chandrashekar , and Z henhai Dua n, Limit ‐ ing Pa t h Exploration in BGP INFOCOM 2005. 24th Annual Joint Conference of the IEEE Compute r and Com munications Socie ‐ ties. Proceedings IEEE, March 2005 page(s): 2337 ‐ 2348 vol. 4 [5] Dmitri Krioukov , Kevin Fall, KC Claffy , and Arthur Brady “ On Compact Routing for the Internet”, ACM SIGCOMM Computer Communication Review , Vo l u m e 37, Number 3, July 2007, pag ‐ es:42 ‐ 52 [6] Duato, J., Casado, R., Bermudez, A., Quiles, F.J.,” Influence of network size and load on the per formance of reconfiguration protocols”, Network Computing and Applications, 2001. NCA 2001. IEEE International Symposium on, Page(s): 46 –57, 2001. [7] Nick Feamster , David G. Andersen, Hari Balakrishnan, and M. © 2009 Journal of Computing http://sites.google.com/site/jo urnalofcomputing/ JOURNAL OF COMPUTING, VOLUME 1, ISSUE 1, DECEMBER 2009, ISSN: 2151-9617 HTTPS://SITES.GOOGLE.COM/S ITE/JOURNALOFCOMPUTING/ 118 Frans Kaashoek, Measuring the Effec ts of Internet Pa t h Faults on Reactive Routing ., Joint International Conference on Meas ‐ urement and Modeling of Computer Syste ms , Proceedings of the 2003 ACM SIGMETRICS international conference on Meas ‐ urement and modeling of compu ter systems. San Diego, CA, USA, Pa g e s : 126 – 137 Ye a r : 2003 [8] Neil Timothy Spring, Efficient discovery of network topolo gy and routing policy in the Internet, Ph D dissertatio n, University of Wa s h i n g t o n , ye ar 2004, A v ailable from www .cs.umd.edu/~ns pring/papers/nsprin g ‐ thesis [9] Ye h u d a Afe k, Anat Bre mler ‐ Barr , and Shemer Schwarz, Im ‐ proved BGP Convergence via Ghost Fl ushing, IN FOCOM 2003. Tw e n t y ‐ Second Annua l Jo in t Conference of the IEEE Computer and Communications . IEEE Societies Publication Date: 30 March ‐ 3 April 2003 Vo l u m e : 2, On page(s): 927 ‐ 937 vol. 2 [10] Dan Pei, Matt Azuma, Dan Massey , and Lixia Zhang, im prov ‐ ing BGP convergence thr ough r oot cause notifi cation. El sevier Computer Networks, Vo l u m e 48, Issue 2, 6 June 2005, Pa g e s 175 ‐ 194. [11] Hongsuda Ta ngmunarunki t, Ramesh Go vindan, S cott Shenk er, and Deborah Estrin, The Impact of Routing Policy on Internet Paths, INFOCO M 2001. T wentieth Annual Joint Conference of the IEEE Computer and Communications Societies. Proceed ‐ ings. IEEE Computer and Communications, Societies. Proceed- ings, IEEE, Publication Date: 2001 pages: 736- 742 vol.2 [12] Timothy G. Griffin, F. Bruce Shepherd, and Gordon Wilfong, The Stable Pa t h s Problem and Interdomain Routin g, IEEE/ACM Tr a n s a c t i o n s on Networking , Vo l u m e 10, Issue 2, Pa g e s : 232 – 243, Ye a r : 2002 [13] George Siganos, Michalis Faloutsos, Improving the Security and Robustness of Internet Routing: What Can We Do To d a y ? A vailable at www .cs.ucr .edu/~siga nos/papers/security.pdf [14] Feng Wa n g , Jian Qiu, “ On Understanding T ransient Interdo ‐ main Routi ng Failures, IEEE/ACM TR ANSACTIONS ON NETWORKING, VOL. 17, NO. 3, JUNE 2009 [15] Ve r n Pa x s o n , End ‐ to ‐ En d Routing Behavior in the Internet, IEEE/ACM TRANSACTIONS ON NETWORKING, VOL. 5, NO. 5, OCTOBER 1997 [16] Bassa m Halabi, Internet Routi ng Architectures, 2 nd Editio n, Cisco Press. [17] Y. Rekhter, T. Li, Border Gatew ay Protocol 4, Rfc 1771, SRI Network Information Center, July 1995. [18] Y. Rekhter , T. Li, S. Hares, Border Gatew ay Protocol 4, A vailable from October 2003. [19]http://www.cisco.com/en/US/docs/ios/iproute/configuration/gui ‐ de Dr Shishir Kumar is currently working as Associate Professor in the department of Computer Science and Engineering; He is also Head of the Computer Science and Engieering Department, in Jaypee Institute of Engineering and Technology, Guna, Madh ya Pradesh, India. His academic qualification is Ph. D. (Computer Science) with more than 12 years of expe rience in rsearch and teaching. Mahesh Kumar is currently a lecturer in the department of Computer Science and Engineering in Jaypee Institute of Engineering and T echnology , Guna, Madhya Pradesh, India, He has received is M. T ech. degree in Inform ation T echnolog y from Punjabi Un iversity , Patiala. He has more than 8 years of experience of managing computer networks (Cisco routers, switches, firewalls), and teaching. © 2009 Journal of Computing http://sites.google.com/site/jo urnalofcomputing/
Original Paper
Loading high-quality paper...
Comments & Academic Discussion
Loading comments...
Leave a Comment