K. Sklower. "A Tree-Based Packet Routing Table for Berkeley Unix", Computer System Research Group, EECS Department, Computer Science Division, University of California at Berkeley, Jul 1991.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....an entry consumes a non negligible volume of the total trac, any approximation will be able to detect it. 9 To limit memory use and search time with variable length keys, we employ a Patricia tree. Patricia has been employed in the BSD kernel for the internal representation of the routing table [23], and its performance characteristics are well understood. It is suitable to handle 32 bit IPv4 addresses and 128 bit IPv6 addresses. Patricia is a full binary radix tree. All internal nodes have exactly two children so that when the number of leaf nodes is N , the number of internal nodes is (N ....

Keith Sklower. A tree-based packet routing table for berkeley UNIX. In USENIX Winter Conference, Dallas, Texas, January 1991.

....to the COA. Any number of mvif host entries may be 2 This is a host route; that is, a 32 route to a particular MN IP address. 7 installed in the routing table. We felt this mechanism allowed us to take direct advantage of the BSD routing radix algorithm mechanism, which is very efficient [16]. Logically such a route looks like this: to MN IP via COA IP thru mvif device Foreign agents are relatively simple (barring fundamental kernel modifications) They use the mvif device to decapsulate IPIP packets, and simply install (or remove) link layer routes for local MNs. A few kernel ....

K. Sklower, "A Tree-Based Packet Routing Table for Berkeley UNIX", Proceedings of the

....the program itself is short and can be entirely stored in any reasonable instruction cache, while the implementation of the data structures for the address cache is a primary concern. The routing table lookup code was extracted from the FreeBSD [9] 4. 3 Release kernel, which uses a radix tree [10] data structure to organize the routing table. The routing table itself was obtained from the same University of Alberta router at which the real UofA trace was collected. The simulation tool used was SimpleScalar[11] 0 0.02 0.04 0.06 0.08 0.1 0.12 0 5 10 15 20 25 30 line size real ....

K. Sklower. A tree-based packet routing table for berkeley unix. In USENIX Winter '91, Dallas, TX, USA., 1991.

....across subnets within its domain. Maintaining these per host entries need not be a 354 R. C aceres, V.N. Padmanabhan Fast and scalable wireless handoffs problem because suitable data structures exist that allow modern routers to handle several tens of thousands of entries efficiently (e.g. [33]) The need for such hierarchies of foreign agents is a topic of discussion in the Mobile IP Working Group of the Internet Engineering Task Force. Similar hierarchies have been recently and independently proposed by Johnson [16] and Perkins [23] An earlier proposal for hierarchical location ....

K. Sklower, A tree-based packet routing table for Berkeley Unix, in: Proc.

....with which to add or delete static rules. All such rules must be present in the source code and the module compiled and reinstalled each time a rule is changed. When large rule bases are in use, the efficiency of fw could be improved by replacing the linear table structure with a Patricia tree[8]. It could also be argued that the rule matching logic should be extended to support protocol numbers, port numbers, and hard connection state. We would argue that if this were done, then fw would no longer be fw. It would be yasifw. Source code for fw is available from the author via e mail ....

Sklower, K. "A Tree-Based Packet Routing Table for Berkeley Unix", Proc.

....matching a destination IP address. Thus, standard techniques for exact matching, such as perfect hashing, binary search, and standard Content Addressable Memories (CAM) cannot directly be used for Internet address lookups. Also, the most widely used algorithm for IP lookups, BSD Patricia Tries [Sklower 1993], has poor performance. Prefix matching in Internet routers was introduced in the early 1990s, when it was foreseen that the number of endpoints and the amount of routing information would grow enormously. At that time, only address classes A, B, and C existed, giving individual sites either 24, ....

.... natural solution for exact matching, a natural question is: Why can t we modify hashing to do best matching prefix However, for several years now, it was considered not to be apparent how to accommodate hierarchies while using hashing, other than rehashing for each level of hierarchy possible [Sklower 1993]. Our paper describes a novel algorithmic solution to longest prefix match, using binary search over hash tables organized by the length of the prefix. Our solution requires a worst case of log W hash lookups, with W being the length of the address in bits. Thus, for the current Internet protocol ....

[Article contains additional citation context not shown here]

SKLOWER, K. 1993. A tree-based packet routing table for Berkeley Unix. Technical report, University of California, Berkeley. Also at http://www.cs.berkeley.edu/ sklower/routing.ps.

....matching a destination IP address. Thus, standard techniques for exact matching, such as perfect hashing, binary search, and standard Content Addressable Memories (CAM) cannot directly be used for Internet address lookups. Also, the most widely used algorithm for IP lookups, BSD Patricia Tries [Sklower 1993], has poor performance. Prefix matching in Internet routers was introduced in the early 1990s, when it was foreseen that the number of endpoints and the amount of routing information would grow enormously. At that time, only address classes A, B, and C existed, giving individual sites either 24, ....

.... natural solution for exact matching, a natural question is: Why can t we modify hashing to do best matching prefix However, for several years now, it was considered not to be apparent how to accommodate hierarchies while using hashing, other than rehashing for each level of hierarchy possible [Sklower 1993]. Our paper describes a novel algorithmic solution to longest prefix match, using binary search over hash tables organized by the length of the prefix. Our solution requires a worst case of hash lookups, with being the length of the address in bits. Thus, for the current Internet ....

[Article contains additional citation context not shown here]

SKLOWER, K. 1993. A tree-based packet routing table for Berkeley Unix. Technical report, University of California, Berkeley. Also at http://www.cs.berkeley.edu/ sklower/routing.ps.

....IP destination address as key. This lookup returns the best matching routing table entry, which tells the router which interface to forward the packet out of and the IP address of the packet s next hop 7 The first approaches for longest prefix matching used radix trees or modified Patricia trees [13][14] combined with hash tables, Patricia stands for Practical Algorithm to Retrieve Information Coded in Alphanumeric ) These trees are binary trees, whereby the tree traversal depends on a sequence of single bit comparisons in the key, the destination IP address. These lookup algorithms have ....

. K. Sklower, "A Tree-Based Packet Routing Table for Berkeley Unix," USENIX, Winter'91, Dallas, TX, 1991.

....an INR that interact with the name tree. One of these is the code that handles the resolution of name speci ers, and the other is the code that is responsible for sharing information about name speci ers with other INRs. Thus, the function of a name tree is similar to that of an IP routing table [35] or name server cache [38] For resolving name speci ers, the name tree supports a Lookup Name operation, which returns the information associated with a name speci er in the name tree. Since the rate of lookups an INR must perform will grow as the system becomes more heavily used, lookups must ....

K. Sklower. A Tree-Based Packet Routing Table for Berkeley Unix. Technical report, University of California, Berkeley, 1993.

....this work, and a brief outline of on going work. II. RELATED WORK The most popular search data structure for the longest prefix string match problem is the bit by bit Patricia trie [1] A similar technique called reduced radix tree [11] has been implemented in the 4. 3 version of Berkeley Unix [2]. Pei et al. 3] compared the VLSI implementation cost and the performance of content addressable memory (CAM) based and trie based solutions to the routing table lookup problem. McAuley and Francis [4] proposed several fast routing table lookup solutions for variable length hierarchical network ....

.... the physical addresses that get mapped to the first 4KBytes of the cache would be those that lie in physical pages whose page numbers are a multiple of 3 0 0 0 0 bits [11, 5] bits [31, 12] OutPort Tag 32 bytes 384 sets 512 sets 32 bits L1 Cache Translation Address Cache Lookup bits [4, 2] Address Address Physical ignored bits [1, 0] Software Compare Destination IP Address (DA) Virtual HAC Fig. 1. The data flow of HAC lookup. Here we assume that L1 cache is direct mapped with 512 32 byte cache sets, among which 128 cache sets are reserved for HAC. Each HAC entry is 4 ....

Sklower, K., A tree-based packet routing table for Berkeley UNIX, Proceedings of the Winter 1991 USENIX Conference, p. 93-103, Dallas, TX, USA 21-25 Jan. 1991.

....timer, but they are ignored in this paper. 2 Related Work The most popular search data structure for the longest prefix string match problem is the bitby bit Patricia trie [1] A similar technique called reduced radix tree [11] has been implemented in the 4. 3 version of Berkeley Unix [2]. Pei et al. 3] compared the VLSI implementation cost and the performance of content addressable memory (CAM) based and trie based solutions to the routing table lookup problem. McAuley and Francis [4] proposed several fast routing table lookup solutions for variable length hierarchical network ....

....X ranging from the n th bit to the m th bit. 0 0 0 0 0 1 1 1 000000000000111111111111 000 000 111 111 0 0 1 1 0000000000000 0000000000000 1111111111111 1111111111111 0 0 bits [11, 5] bits [31, 12] OutPort Tag 32 bytes 384 sets 512 sets 32 bits L1 Cache Translation Address Cache Lookup bits [4, 2] Address Address Physical ignored bits [1, 0] Software Compare Destination IP Address (DA) Virtual HAC Figure 1: The data flow of HAC lookup. Here we assume that L1 cache is direct mapped with 512 32 byte cache sets, among which 128 cache sets are reserved for HAC. Each HAC entry is 4 bytes ....

Sklower, K., "A tree-based packet routing table for Berkeley UNIX," Proceedings of the Winter 1991 USENIX Conference, p. 93-103, Dallas, TX, USA 21-25 Jan. 1991.

....a contiguous address block to a single route [4] In order to support CIDR and VLSM, routers must provide efficient best match lookup into a large routing table. The 4. 3BSD Reno release introduced the reduced radix tree for routing table instead of the hash based scheme of previous releases [5]. The reduced radix tree is general enough to support variable length addresses such as OSI NSAP addresses and is powerful enough to perform best match lookup even if non continuous subnet masks were allowed. The lookup and maintenance algorithms are, however, complicated due to support of ....

....section 4 and give a conclusion in the last. Throughout this paper, we describe a route in (address, mask) notation as well as address prefix notation in address masklen. For example, 133.5.16.0, 0xffffff00) is equivalent to 133.5.16.0 24. 2 Radix This section first reviews Radix described in [5]. Three examples are given for comprehension of its lookup algorithm. Then we claim to complicated lookup and tree maintenance procedure of Radix. 2.1 Radix tree structure Best match lookup, sometimes called longest match, is to find a matched entry in which the number of bit set in the mask is ....

K. Sklower, "A Tree--Based Packet Routing Table for Berkeley Unix", USENIX Winter, pp. 93 -- 99, 1991.

....of this work, and a brief outline of on going work. 2 Related Work The most popular search data structure for the longest prefix string match problem is the bitby bit Patricia trie [1] A similar technique called reduced radix tree [11] has been implemented in the 4. 3 version of Berkeley Unix [2]. Pei et al. 3] compared the VLSI implementation cost and the performance of content addressable memory (CAM) based and trie based solutions to the routing table lookup problem. McAuley and Francis [4] proposed several fast routing table lookup solutions for variable length hierarchical network ....

....X ranging from the n th bit to the m th bit. 0 0 0 0 0 1 1 1 000000000000111111111111 000 000 111 111 0 0 1 1 0000000000000 0000000000000 1111111111111 1111111111111 0 0 bits [11, 5] bits [31, 12] OutPort Tag 32 bytes 384 sets 512 sets 32 bits L1 Cache Translation Address Cache Lookup bits [4, 2] Address Address Physical ignored bits [1, 0] Software Compare Destination IP Address (DA) Virtual HAC Figure 1: The data flow of HAC lookup. Here we assume that L1 cache is direct mapped with 512 32 byte cache sets, among which 128 cache sets are reserved for HAC. Each HAC entry is 4 bytes ....

Sklower, K., "A tree-based packet routing table for Berkeley UNIX," Proceedings of the Winter 1991 USENIX Conference, p. 93-103, Dallas, TX, USA 21-25 Jan. 1991.

....for inbound and outbound packets. The Key Engine also communicates with user level key management programs so that key management may be implemented properly. The relationship between the key engine and user level key management programs is similar to the relationship between the routing socket[Skl91] and programs such as gated(8) 3.2 Security Processing Structure The authentication processing function is split into three major parts. The first, a keyed message digest function, is selected on a per association basis through an algorithm switch that calls the appropriate computation ....

....This extension to Sockets provides a generic interface between security association management applications, such as a Photuris daemon [KS95] and the kernel s network security data structures. PAM95] This new generic key management interface is modeled upon the existing routing socket, PF ROUTE. Skl91] This enhancement permits the key management system to be completely decoupled from the IP security implementation. Multiple key management schemes can be supported concurrently if desired. It also will make it easy to change from one key management algorithm or protocol to a new key management ....

Keith Sklower. A Tree-Based Packet Routing Table for Berkeley UNIX. In Proceedings of the Winter '91 USENIX Conference, Dallas, TX, January 1991. USENIX Association.

....per mobile host routing entries and update them whenever a mobile moves across subnets within its domain. Maintaining these perhost entries need not be a problem because suitable data structures exist that allow modern routers to handle several tens of thousands of entries efficiently (e.g. [23]) The need for such hierarchies of foreign agents is now a topic of discussion in the Mobile IP Working Group of the Internet Engineering Task Force. Similar hierarchies have been recently and independently proposed by Johnson and Perkins [13] 19] The third level in the hierarchy handles global ....

K. Sklower, "A Tree-Based Packet Routing Table for Berkeley Unix," Proc. of the 1991 Winter USENIX Technical Conference, January 1991.

....application programmers, such as the sockets application programming interface (API) LMKQ89] The sockets interface allows applications in BSD to communicate with other applications, or sometimes, even with the operating system itself. One of the recent developments in BSD was the routing socket [Skl91], which allows a privileged application to alter a node s network routing tables. This abstraction allows a BSD system to use an appropriate routing protocol, without requiring changes inside the kernel. Instead, routing protocols are implemented in userspace daemons, such as routed or gated. An ....

....on another network communication. Conclusion Presented here is an application programming interface (API) and surrounding infrastructure to allow modular and relatively easy implementation of key management strategies. It uses concepts and abstractions introduced in various works [LMKQ86] [Skl91] [Atk95a] and ties them together so that good abstractions to both key management strategies and networking protocol implementation. A sample implementation of PF KEY is available in the NRL IPv6 IPsec distribution. The distribution can be obtained by U. S. citizens at ftp: ftp.c2.org, or by ....

Sklower, K., "A Tree-Based Packet Routing Table for Berkeley UNIX," Proceedings of the 1991 Winter USENIX Conference, pp. 93-99.

....per mobile host routing entries and update them whenever a mobile moves across subnets within its domain. Maintaining these per host entries need not be a problem because suitable data structures exist that allow modern routers to handle several tens of thousands of entries efficiently (e.g. [33]) The need for such hierarchies of foreign agents is a topic of discussion in the Mobile IP Working Group of the Internet Engineering Task Force. Similar hierarchies have been recently and independently proposed by Johnson [16] and Perkins [23] An earlier proposal for hierarchical location ....

K. Sklower, "A Tree-Based Packet Routing Table for Berkeley Unix," Proc. of the 1991 Winter USENIX Technical Conference, January 1991.

No context found.

K. Sklower. "A Tree-Based Packet Routing Table for Berkeley Unix", Computer System Research Group, EECS Department, Computer Science Division, University of California at Berkeley, Jul 1991.

No context found.

K. Sklower, A Tree-Based Packet Routing Table for Berkeley UNIX, Proc. of the Winter 1991.

No context found.

K. Sklower. "A Tree-Based Packet Routing Table for Berkeley Unix", Computer System Research Group, EECS Department, Computer Science Division, University of California at Berkeley, Jul 1991.

No context found.

Keith Sklower, A Tree-Based Packet Routing Table for Berkeley Unix, USENIX Winter 1991: 93-104

No context found.

K. Sklower (1991), A Tree-Based Packet Routing Table for Berkeley Unix. In Proceedings of the 1991 Winter USENIX Technical Conference.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC