Multi-layer Structure of Data Center Based on Steiner Triple System
|
|
- Jean Bridges
- 8 years ago
- Views:
Transcription
1 Journal of Computational Information Systems 9: 11 (2013) Available at Multi-layer Structure of Data Center Based on Steiner Triple System Jianfei ZHANG 1, Zhiyi FANG 1, S. Q. ZHENG 2, Guannan QU 1, 1 College of Computer Science and Technology, Jilin University, Changchun , China 2 Department of Computer Science, University of Texas at Dallas, Richardson TX 75083, USA Abstract Topology structure is considered as a primary factor which largely influences the performance of data center (DC). There are lots of results which are deployed graph theories or block design theories. In this paper, we develop an innovate DC topology structure named Multi-layer Steiner Triple System, abbr. MST S. We will prove that MSTS is able to be scaled in faster than double exponential speed and with short diameter. This significant property makes MSTS be built with large number of commercial connection equipments without sacrificing communication performance. Keywords: Data Center; Network Topology; Steiner Triple System; Scalable 1 Introduction In recent years, as Internet getting popular, online application services, such as on-line games, e-businesses, on-line videos, SNS, are becoming a hot spot. All of these services are running around a core of data. Enterprises build their own data center (DC) for storing and processing huge amount of data. With data increasing, data centers have to be extended. It is no doubt that the topology structure of data center is precisely the key point to meet this demands. A well designed topology structure is able to connect as many computation units as possible, with high scalability and short diameter. These requirements raise challenges of designing DC topology structure. Numbers of works were proposed. Bus interconnection network [1] is a networks that the buses provide communication medium between computation units. The solutions of Bus networks are reduce to solving hypergraph. Since it is a point-to-point networks, when it is scaled, its diameter increases rapidly. Hypercube [2] structure C n is a multi-dimension structure which includes 2 n computation units where n is the number of the dimensions and the diameter of the interconnection networks. But in a C n, each node has to connect with n other nodes. Every node is required to deploy n communication ports, such that it is not suitable for a large scale Corresponding author. address: qu.guannan@hotmail.com (Guannan QU) / Copyright 2013 Binary Information Press DOI: /jcis6123 June 1, 2013
2 4372 J. Zhang et al. /Journal of Computational Information Systems 9: 11 (2013) networks. Complete graph structure K n includes n computation units, and these units link to each other. In this way, the distance between any pair of computation units is exactly one. However, every computation unit has to deploy (n 1) communication port. For K n with large size n, n! connection wires are hard to be implemented. N-star structure also faces the scalable issue. In N-star structure S n [3, 4], n! computation units by n!(n 1)/2 switch equipments. DCell structure [5, 6] is a novel data center topology structure which is constructed with mini-switches. It can be scaled in doubly exponential speed as the node degree increase and keep small diameter. In this paper, we present an innovate topology structure called Multi-layer Steiner Triple System (MST S). We will prove that, the number of computation units connected by MST S increases even faster than doubly exponentially, however, the diameter of a MST S l is only 2 l+1 1. Comparing to DCell structure, with the same diameter, MST S can connect times number of computation units. It improves the performance of the communication significantly. And computation units can be connected by low-end switch equipments. It largely reduces the implementation cost. 2 Preliminaries 2.1 Block design Combinatorial Design [7] is a way of selecting subsets from a finite set to meet specific conditions, and the conditions may involve set members, set intersections and so on. Block Design is included in Combinatorial Design as one of important parts. A Design is defined on a finite set X = {x 1, x 2,, x v }, where element x i is called Treatment or Variety. And all treatments are partition into a non-empty collection of subsets of X. The collection of subsets is like B = {B 1, B 2,, B b }, where B i is called a Block. The number of blocks in which a treatment appears is called the valence of that treatment. The number of blocks in which any pair of treatments appears is called the covalence of that pair of treatments. A Regular Design is a Block Design which satisfies two conditions : I: all blocks have the same size of k, k 2; II: all treatments have the same valence, such that every treatment appears in the same number r of blocks, where r is called replication number. From above definitions, we have bk = vr, (1) where v is the number treatments, b is the number of blocks, r is the replication number and k is the size of block. If all v treatments appear in the same block of a design, such block is called Complete. But if there is at least one incomplete block, we call the design is Incomplete. In an incomplete block design, assuming x and y are two different treatments, their covalence is represented as λ xy. We say a block design is balanced, if each pair of treatments has the same covalence, and they appear in the same number λ of blocks. λ is called the index of the design. The block design which has been studied more frequently is Balanced Incomplete Block Design, abbr. BIBD. It is represented as a five-tuple (v, b, r, k, λ). And in a (v, b, r, k, λ)-bibd, we have
3 J. Zhang et al. /Journal of Computational Information Systems 9: 11 (2013) r(k 1) = λ(v 1). (2) 2.2 Steiner triple system Steiner Systems is a type of block design with λ = 1. A steiner systems, denoted by S(t, k, v), is based on a set V with cardinality v. and V is partitioned into a subsets collection B who has size k. Each t-element subset of V is contained exactly in one member of B. In terms of hypergraph, each element in V is a vertex, and each k-subset in B is a hyperedge, S(t, k, v) is a uniform hypergraph who has v nodes, and each edge contains k nodes. S(2, 3, v), is a special steiner systems with t = 2 and k = 3. We use ST S(v) to present a steiner triple system of cardinality v. A ST S(v) is built on a elements set of v. The v elements are partition into b blocks and each block contains 3 elements. Each two elements appears in exactly one of b blocks. A S(2, 3, v) has C(v, 2) 2-subsets and C(v, 3) 3-subsets. Every 3-subset contains C(3, 2) = 3 2-subsets. Thus a solution to S(2, 3, v) has b = C(v, 2)/C(3, 2) 3-subsets. Resolving the equation, we get: b = v(v 1). (3) 6 r is the number of 3-subsets that an element appears. Apparently, according to (1), we have rv = 3b, i.e. Replacing the b in Equation (4) by Equation (3), we have r = 3b v. (4) and r = v 1 2, (5) v = 2r + 1. (6) The following is an example of ST S(9) on set V = {1, 2, 3, 4, 5, 6, 7, 8, 9}, ST S(9) = {{1, 2, 3}, {1, 4, 5}, {1, 6, 7}, {1, 8, 9}, {2, 4, 6}, {2, 5, 8}, {2, 7, 9}, {3, 4, 9}, {3, 5, 7}, {3, 6, 8}, {4, 7, 8}, {5, 6, 9}}, where ST S(9) is based on a elements set of 9. The elements are partition into b = 12 blocks, and each elements appears in r = 4 blocks. Any pair of elements presents in one block. The necessary and sufficient condition [8] for the existence of ST S(n) is n 1 or 3 ( mod 6 ). (7)
4 4374 J. Zhang et al. /Journal of Computational Information Systems 9: 11 (2013) And we say that an positive integer n is admissible if and only if n is congruent to 1 or 3 mod 6. Apparently, the hypergraph representation of ST S(v) is a regular, uniform and linear hypergraph. A ST S(v) hypergraph can connect v vertices with degree of v 1 v(v 1). It has hyperedges with 2 6 rank 3, and any two different nodes in hypergraph are connected by exactly one common edge. 3 Multi-layers Structure Based on STS In this article, we consider a hyper networks modeled by ST S(n) as a basic interconnection networks, and construct a multi-layers interconnection hypernetworks recursively by STS-hyper networks, denoted by MST S. The MST S solution is an excellent hypernetworks with advantages of high scalability and short communicating path. 3.1 M ST S structure The computation units in MST S have multiple network ports and are connected by low-end connection equipments. In M ST S computation units communicate with each other via commercial connection equipments, such as small scale switches. The whole hypernetworks is built recursively, such that high-level MST S l is constructed by low-level MST S l 1. The MST S 0 is constructed by n computation units and a n-ports switch, all computation units connect to the switch directly. MST S 0 is the basic block to build larger MST S. Considering a MST S 0 as a single unit, we construct level-1 MST S 1 using v MST S 0 s. And in a MST S 1, the rules of connections among all MST S 0 are refer to the solution of ST S(v). Since one MST S 0 is as an element of a ST S(v), each element presents in r = n 3-subsets. By the Equation (6), MST S 0 consists of 2r + 1 MST S 0 s. Certainly, the value of 2r + 1 should satisfy the condition of Equation (7). We will prove this later. According to the Equation (3), all 2r + 1 MST S 0 s are connected via v(v 1) 3-ports switches. We label MST S 6 0 with i where i [1, v]. For every 3-subset {i, j, k} in solution of ST S(v), we pick one computation unit which has not connect to any level-1 switches from i, j and k MST S 0, and connect them by one 3-port switch. (1,2,4) (1,3,7) (1,5,6) (2,3,5) (2,6,7) (3,4,6) (4,5,7) Fig. 1: A MST S 1 hypernetworks with n=3.
5 J. Zhang et al. /Journal of Computational Information Systems 9: 11 (2013) An example of MST S 1 with n = 3 is shown in Fig 1. In MST S 1, we treat MST S 0 as a virtual unit, and all MST S 0 s form hypernetworks according to the solution of ST S(v). The level-2 and higher MST S l are constructed in the same way. If the MST S l 1 has been constructed and each MST S l 1 has r l 1 computation units, we can build a MST S l with v l = 2r l MST S l 1 s. As construction of MST S 1, we treat MST S l 1 as a virtual unit, and all MST S l 1 s form a solution of ST S(v l ). Let r l denote the total amount of computation units of MST S l, we have and v l = 2r l 1 + 1, (8) r l = v l r l 1. (9) for l 1. The basic block MST S 0 is different with v 0 = 1 and r 0 = n, where n is the number of computation units in MST S 0. Since in different level i, v i, the number of virtual node MST S i 1, are different. In order to make sure that for i 1, v i always satisfies the condition (7), we have follow theorem. Theorem 1 If n 0 or 1 (mod 3), then ST S(V l ) exists for l 1. Proof According to the necessary and sufficient condition for existence of a ST S(v), there exists a solution of ST S(v), if and only if n 1or3 ( mod 6 ), where n is the number of connection equipment ports in MST S 0. If n 0 or 1 ( mod 3 ), then r 0 = n 0 or 1 ( mod 3 ). We consider following two cases: Case I: suppose r 0 = 3m, m = 1, 2,, n,. According to (6), we have v 1 = 2r 0 + 1, thus v 1 = 6m + 1, m = 1, 2,, n,. Obviously, v 1 satisfies the condition (7). According to Equation (9), r 1 = v 1 r 0, such that r 1 = 3(6m 2 + m). Apparently, r 1 0 (mod 3). Let m 1 = 6m 2 + m, we have r 1 = 3m 1. And v 2 = 6m satisfies the condition (7). In the same way, for i 1, v i = 6m i satisfies the condition (7), where m i 1 = 6m 2 i 2 + m i 2. Case II: suppose r 0 = 3m + 1, m = 1, 2,, n,. According to Equation (6), we have v 1 = 2r 0 +1, thus we have v 1 = 6m+3, m = 1, 2,, n,. Therefore v 1 satisfies the condition (7). From Equation (9), we have r 1 = v 1 r 0, where r 1 = 3(6m 2 + 5m + 1). Let m 1 = 6m 2 + 5m + 1, we have r 1 = 3m 1. v 2 = 6m satisfies the condition (7). As same as case I, for i 1, v i satisfies the condition (7). Concluding above two cases, theorem 1 is proved. 3.2 Properties of MSTS Firstly, we investigate how many computation units a MST S l can be included. Theorem 2 2 l (n )2l 1 4 < r l < 2 l (n + 1) 2l 1.
6 4376 J. Zhang et al. /Journal of Computational Information Systems 9: 11 (2013) Proof From Equation (9) and (8), we have r l = r l 1 (2r l 1 + 1). Thus, we get that is r l = 2(r l )2 1 8 > 2(r l )2 1 4, Solving above formulation iteratively, we get r l > 2(r l )2. and r l > 2 l (n )2l 1 4, (10) Equation (11) can be simplified as r l = 2(r l 1 + 1) 2 3r l 1 2 < 2(r l 1 + 1) 2 1. (11) Solve it iteratively, we get r l + 1 < 2(r l 1 + 1) 2. From Equation (10) and Equation (12), Theorem 2 is proved. r l < 2 l (n + 1) 2l 1. (12) From theorem 2, we find out that the number of computation units included in a MST S l increases even more rapidly than doubly exponentially. Table 1 shows some examples. The first column of Table 1 lists the number of computation units in MST S 0 n. The other three columns are value of l, which is the number of layers of whole MST S structure from 1 to 3 respectively. And the numbers in other cells are the number of computation units connected by a MST S with corresponding n and l connects. For instance, when n = 6 and l = 3, the number of computation units in a MST S 3 is more than 299-million. With the same n, as layer number increasing, the amount of computation units scales hundreds even thousands times. Not only the amount is increasing, but the increasing speed is accelerating. In columns, a small addition to n leads to big expanding of amount of computation units in any layer. Although every computation unit has to deploy more than one communication port, compared to the size of the whole network being extended to, the number of links that one computation unit has is still small. The diameter of a networks is defined as the length of longest shortest path between every pair of nodes. The length of communication path between any pair of computation units in a
7 J. Zhang et al. /Journal of Computational Information Systems 9: 11 (2013) Table 1: Computation Unit Number in MST S l n l = 1 l = 2 l = , ,941, ,653 6,880,407,471 data center is not longer than the diameter. It means that the diameter is a upper bound of the path length. Small diameter provides efficient communication. The following theorem proves the diameter of the MST S l networks. Theorem 3 The diameter of a l-level MST S l is 2 l+1 1. Proof As we know, for 1 i < l, any two different MST S i s are connected by one connection equipment. Without loss of generality, let s and d be source and destination computation units respectively, where s and d belong to different MST S l 1. We use n 1 and n 2 to denote the medium computation units which are in the same MST S l 1 as s and d respectively. In the case of longest path, s n 1 and d n 2. The problem is reduced to to find the sub-paths between s to n 1 and n 2 to d. The length is Solving Equation (13) recursively, we get Theorem 3 is proved. Len l = 2Len l (13) Len l = 2 l+1 1. Diameter is one of important factors for designing a data center topology structure. From theorem 2, we have r l > 2 l n 2l. Such that 2 l < log r l n logn 2l < log r l n. The diameter of MST S is less than 2 log r l n 1. As a comparison, a fat-tree structure has a diameter of 2 log N 2, where N is the number of nodes. Let D ft and D msts be the diameter of fat-trees and MST S respectively. With same number of nodes, r l = N, we have D ft /D msts log n 2. According to theorem 1, n = 3m or 3m + 1 for m = 1, 2, 3,, n,, thus [D ft /D msts ] log 3m 2 or [D ft /D msts ] log 3m+1 2. A hypercube structure has a diameter of log N 2, with N is the number of nodes. And in the same way, we have a similar result where D hc is the diameter of a hypercube. [D hc /D msts ] log 3m 2 /2 or [D hc /D msts ] log 3m+1 2 /2. From Table 2, no matter what the value of n is, to connect same number of nodes, the diameter of Fat-trees structure is multiple times diameter of MST S. And the same result is held with
8 4378 J. Zhang et al. /Journal of Computational Information Systems 9: 11 (2013) Table 2: Diameter Ratio m 3m [D ft /D msts ] [D hc /D msts ] 3m + 1 [D ft /D msts ] [D hc /D msts ] 1 3 log 3 2 (log 3 2)/2 4 log 4 2 (log 4 2)/2 2 6 log 6 2 (log 6 2)/2 7 log 7 2 (log 7 2)/2 3 9 log 9 2 (log 9 2)/2 10 log 10 2 (log 10 2 )/2 hypercube structure, when n > 4. Comparing to a DCell structure whose diameter is 2 k+1 1, k is the number of levels of a DCell, MST S can connect 2 l times computation units than DCell with the same diameter. In another words, diameter of a MST S is shorter than DCell with the same scale. 4 Concluding Remarks In this paper, we presented an innovative topology networks of data center. called Multi-layer Steiner Triple System (MST S). The construction of MST S structure is based on the theory of Steiner Triple System which is a type of block design. In M ST S structure, computation units are connected by low-end commercial connection equipments. The level-0 consists of one n-ports connection equipments and n computation units. And in each higher level, MST S l consists of MST S l 1 s and organizes them according to solution of a ST S. Each computation unit has one level-i links, which connects with two other computation units in different MST S i 1 s. We proved that M ST S has outstanding properties on scalability and communication path. By selecting n carefully, an available solution to construct the M ST S for any layer can be guaranteed. The amount of computation units scales by speed of times double exponent, and the diameter of a MST S is also competitive. References [1] J-C. Bermond, F.O. Ergincan, Bus Interconnection Networks, Discrete Applied Mathematics, issue 68, pp. 1-15, [2] Y. Saad, M.H. Schultz, Topological Properties of Hypercubes, IEEE Trans. on Computers, issue 7, vol. 37, pp , [3] S.G. Akl, T. Wolff, Efficient sorting on the star graph interconnection network, Telecommunication System, issue 10, pp. 3-20, [4] D.K. Saikia, R.K. Sen, Two ranking schemes for efficient computation on the star interconnection network, IEEE Trans. on Parallel and Distributed Systems, vol 7, no.4, pp , [5] C.Guo, H.Wu, K.Tan, L.Shi, Y.Zhang, S.Liu, DCell: A Scalable and Fault-Tolerant Network Structure for Data Centers, Proc. of the ACM SIGCOMM 2008 conference on Data communication, August 17-22, [6] M. Kliegl, J.Lee, J. Li, X. Zhang, C. Guo, D. Rincon, Generalized DCell Structure for Load- Balanced data center networks, proc. of Infocom 2010, [7] W.D. Wallis, Combinatorial Designs, CRC Press1988. [8] Stephen J. Hartley, Aaron H. Konstam, Using Genetic Algorithms to Generate Steiner Triple Systems, Proc. of the 21st Annual ACM Computer Science Conference, 1993.
Generalized DCell Structure for Load-Balanced Data Center Networks
Generalized DCell Structure for Load-Balanced Data Center Networks Markus Kliegl, Jason Lee,JunLi, Xinchao Zhang, Chuanxiong Guo,DavidRincón Swarthmore College, Duke University, Fudan University, Shanghai
More informationSystem Interconnect Architectures. Goals and Analysis. Network Properties and Routing. Terminology - 2. Terminology - 1
System Interconnect Architectures CSCI 8150 Advanced Computer Architecture Hwang, Chapter 2 Program and Network Properties 2.4 System Interconnect Architectures Direct networks for static connections Indirect
More informationThe Goldberg Rao Algorithm for the Maximum Flow Problem
The Goldberg Rao Algorithm for the Maximum Flow Problem COS 528 class notes October 18, 2006 Scribe: Dávid Papp Main idea: use of the blocking flow paradigm to achieve essentially O(min{m 2/3, n 1/2 }
More informationMEGA data centers have emerged as infrastructures for
IEEE TRANSACTIONS ON COMPUTERS, VOL. 62, NO. 7, JULY 2013 1303 Expandable and Cost-Effective Network Structures for Data Centers Using Dual-Port Servers Deke Guo, Member, IEEE, Tao Chen, Member, IEEE,
More informationHyper Node Torus: A New Interconnection Network for High Speed Packet Processors
2011 International Symposium on Computer Networks and Distributed Systems (CNDS), February 23-24, 2011 Hyper Node Torus: A New Interconnection Network for High Speed Packet Processors Atefeh Khosravi,
More informationWhy? A central concept in Computer Science. Algorithms are ubiquitous.
Analysis of Algorithms: A Brief Introduction Why? A central concept in Computer Science. Algorithms are ubiquitous. Using the Internet (sending email, transferring files, use of search engines, online
More informationEvery tree contains a large induced subgraph with all degrees odd
Every tree contains a large induced subgraph with all degrees odd A.J. Radcliffe Carnegie Mellon University, Pittsburgh, PA A.D. Scott Department of Pure Mathematics and Mathematical Statistics University
More information102 IEEE/ACM TRANSACTIONS ON NETWORKING, VOL. 19, NO. 1, FEBRUARY 2011
102 IEEE/ACM TRANSACTIONS ON NETWORKING, VOL. 19, NO. 1, FEBRUARY 2011 Scalable and Cost-Effective Interconnection of Data-Center Servers Using Dual Server Ports Dan Li, Member, IEEE, Chuanxiong Guo, Haitao
More informationData Structures and Algorithms Written Examination
Data Structures and Algorithms Written Examination 22 February 2013 FIRST NAME STUDENT NUMBER LAST NAME SIGNATURE Instructions for students: Write First Name, Last Name, Student Number and Signature where
More informationLoad Balancing and Switch Scheduling
EE384Y Project Final Report Load Balancing and Switch Scheduling Xiangheng Liu Department of Electrical Engineering Stanford University, Stanford CA 94305 Email: liuxh@systems.stanford.edu Abstract Load
More informationDegree Hypergroupoids Associated with Hypergraphs
Filomat 8:1 (014), 119 19 DOI 10.98/FIL1401119F Published by Faculty of Sciences and Mathematics, University of Niš, Serbia Available at: http://www.pmf.ni.ac.rs/filomat Degree Hypergroupoids Associated
More informationFUZZY CLUSTERING ANALYSIS OF DATA MINING: APPLICATION TO AN ACCIDENT MINING SYSTEM
International Journal of Innovative Computing, Information and Control ICIC International c 0 ISSN 34-48 Volume 8, Number 8, August 0 pp. 4 FUZZY CLUSTERING ANALYSIS OF DATA MINING: APPLICATION TO AN ACCIDENT
More informationA RDT-Based Interconnection Network for Scalable Network-on-Chip Designs
A RDT-Based Interconnection Network for Scalable Network-on-Chip Designs ang u, Mei ang, ulu ang, and ingtao Jiang Dept. of Computer Science Nankai University Tianjing, 300071, China yuyang_79@yahoo.com.cn,
More informationTopological Properties
Advanced Computer Architecture Topological Properties Routing Distance: Number of links on route Node degree: Number of channels per node Network diameter: Longest minimum routing distance between any
More informationBig Data Begets Big Database Theory
Big Data Begets Big Database Theory Dan Suciu University of Washington 1 Motivation Industry analysts describe Big Data in terms of three V s: volume, velocity, variety. The data is too big to process
More informationCMSC 858T: Randomized Algorithms Spring 2003 Handout 8: The Local Lemma
CMSC 858T: Randomized Algorithms Spring 2003 Handout 8: The Local Lemma Please Note: The references at the end are given for extra reading if you are interested in exploring these ideas further. You are
More informationOptimal Index Codes for a Class of Multicast Networks with Receiver Side Information
Optimal Index Codes for a Class of Multicast Networks with Receiver Side Information Lawrence Ong School of Electrical Engineering and Computer Science, The University of Newcastle, Australia Email: lawrence.ong@cantab.net
More informationOn the k-path cover problem for cacti
On the k-path cover problem for cacti Zemin Jin and Xueliang Li Center for Combinatorics and LPMC Nankai University Tianjin 300071, P.R. China zeminjin@eyou.com, x.li@eyou.com Abstract In this paper we
More informationApproximated Distributed Minimum Vertex Cover Algorithms for Bounded Degree Graphs
Approximated Distributed Minimum Vertex Cover Algorithms for Bounded Degree Graphs Yong Zhang 1.2, Francis Y.L. Chin 2, and Hing-Fung Ting 2 1 College of Mathematics and Computer Science, Hebei University,
More informationEfficient Recovery of Secrets
Efficient Recovery of Secrets Marcel Fernandez Miguel Soriano, IEEE Senior Member Department of Telematics Engineering. Universitat Politècnica de Catalunya. C/ Jordi Girona 1 i 3. Campus Nord, Mod C3,
More informationA Direct Numerical Method for Observability Analysis
IEEE TRANSACTIONS ON POWER SYSTEMS, VOL 15, NO 2, MAY 2000 625 A Direct Numerical Method for Observability Analysis Bei Gou and Ali Abur, Senior Member, IEEE Abstract This paper presents an algebraic method
More informationA Reliability Analysis of Datacenter Topologies
A Reliability Analysis of Datacenter Topologies Rodrigo S. Couto, Miguel Elias M. Campista, and Luís Henrique M. K. Costa Universidade Federal do Rio de Janeiro - PEE/COPPE/GTA - DEL/POLI Email:{souza,miguel,luish}@gta.ufrj.br
More informationPCube: Improving Power Efficiency in Data Center Networks
PCube: Improving Power Efficiency in Data Center Networks Lei Huang, Qin Jia, Xin Wang Fudan University Shanghai, China 08300240053@fudan.edu.cn 08300240080@fudan.edu.cn xinw@fudan.edu.cn Shuang Yang Stanford
More informationInterconnection Networks Programmierung Paralleler und Verteilter Systeme (PPV)
Interconnection Networks Programmierung Paralleler und Verteilter Systeme (PPV) Sommer 2015 Frank Feinbube, M.Sc., Felix Eberhardt, M.Sc., Prof. Dr. Andreas Polze Interconnection Networks 2 SIMD systems
More informationError Tolerant Address Configuration for Data Center Networks with Malfunctioning Devices
Error Tolerant Address Configuration for Data Center Networks with ing Devices Xingyu Ma,, Chengchen Hu, Kai Chen, Che Zhang, Hongtao Zhang, Kai Zheng, Yan Chen, Xianda Sun, MOE KLINNS lab, Department
More informationTHE DESIGN OF AN EFFICIENT LOAD BALANCING ALGORITHM EMPLOYING BLOCK DESIGN. Ilyong Chung and Yongeun Bae. 1. Introduction
J. Appl. Math. & Computing Vol. 14(2004), No. 1-2, pp. 343-351 THE DESIGN OF AN EFFICIENT LOAD BALANCING ALGORITHM EMPLOYING BLOCK DESIGN Ilyong Chung and Yongeun Bae Abstract. In order to maintain load
More informationmost 4 Mirka Miller 1,2, Guillermo Pineda-Villavicencio 3, The University of Newcastle Callaghan, NSW 2308, Australia University of West Bohemia
Complete catalogue of graphs of maimum degree 3 and defect at most 4 Mirka Miller 1,2, Guillermo Pineda-Villavicencio 3, 1 School of Electrical Engineering and Computer Science The University of Newcastle
More informationAll trees contain a large induced subgraph having all degrees 1 (mod k)
All trees contain a large induced subgraph having all degrees 1 (mod k) David M. Berman, A.J. Radcliffe, A.D. Scott, Hong Wang, and Larry Wargo *Department of Mathematics University of New Orleans New
More informationOutline. NP-completeness. When is a problem easy? When is a problem hard? Today. Euler Circuits
Outline NP-completeness Examples of Easy vs. Hard problems Euler circuit vs. Hamiltonian circuit Shortest Path vs. Longest Path 2-pairs sum vs. general Subset Sum Reducing one problem to another Clique
More informationLarge-Scale Distributed Systems. Datacenter Networks. COMP6511A Spring 2014 HKUST. Lin Gu lingu@ieee.org
Large-Scale Distributed Systems Datacenter Networks COMP6511A Spring 2014 HKUST Lin Gu lingu@ieee.org Datacenter Networking Major Components of a Datacenter Computing hardware (equipment racks) Power supply
More informationTechnology, Kolkata, INDIA, pal.sanjaykumar@gmail.com. sssarma2001@yahoo.com
Sanjay Kumar Pal 1 and Samar Sen Sarma 2 1 Department of Computer Science & Applications, NSHM College of Management & Technology, Kolkata, INDIA, pal.sanjaykumar@gmail.com 2 Department of Computer Science
More informationSPERNER S LEMMA AND BROUWER S FIXED POINT THEOREM
SPERNER S LEMMA AND BROUWER S FIXED POINT THEOREM ALEX WRIGHT 1. Intoduction A fixed point of a function f from a set X into itself is a point x 0 satisfying f(x 0 ) = x 0. Theorems which establish the
More informationBig Data Analytics of Multi-Relationship Online Social Network Based on Multi-Subnet Composited Complex Network
, pp.273-284 http://dx.doi.org/10.14257/ijdta.2015.8.5.24 Big Data Analytics of Multi-Relationship Online Social Network Based on Multi-Subnet Composited Complex Network Gengxin Sun 1, Sheng Bin 2 and
More informationInterconnection Networks. Interconnection Networks. Interconnection networks are used everywhere!
Interconnection Networks Interconnection Networks Interconnection networks are used everywhere! Supercomputers connecting the processors Routers connecting the ports can consider a router as a parallel
More informationBounded Cost Algorithms for Multivalued Consensus Using Binary Consensus Instances
Bounded Cost Algorithms for Multivalued Consensus Using Binary Consensus Instances Jialin Zhang Tsinghua University zhanggl02@mails.tsinghua.edu.cn Wei Chen Microsoft Research Asia weic@microsoft.com Abstract
More informationAutonomous Fault Detection and Recovery System in Large-scale Networks
Autonomous Fault Detection and Recovery System in Large-scale Networks Raheel Ahmed Memon 1, Yeonseung Ryu, Abdul Qadir Rahimoo Abstract In networks like Combat ship data network, the Ethernet is required
More informationAN ALGORITHM FOR DETERMINING WHETHER A GIVEN BINARY MATROID IS GRAPHIC
AN ALGORITHM FOR DETERMINING WHETHER A GIVEN BINARY MATROID IS GRAPHIC W. T. TUTTE. Introduction. In a recent series of papers [l-4] on graphs and matroids I used definitions equivalent to the following.
More informationA Network Flow Approach in Cloud Computing
1 A Network Flow Approach in Cloud Computing Soheil Feizi, Amy Zhang, Muriel Médard RLE at MIT Abstract In this paper, by using network flow principles, we propose algorithms to address various challenges
More informationNetwork (Tree) Topology Inference Based on Prüfer Sequence
Network (Tree) Topology Inference Based on Prüfer Sequence C. Vanniarajan and Kamala Krithivasan Department of Computer Science and Engineering Indian Institute of Technology Madras Chennai 600036 vanniarajanc@hcl.in,
More informationDistributed Computing over Communication Networks: Topology. (with an excursion to P2P)
Distributed Computing over Communication Networks: Topology (with an excursion to P2P) Some administrative comments... There will be a Skript for this part of the lecture. (Same as slides, except for today...
More informationBicolored Shortest Paths in Graphs with Applications to Network Overlay Design
Bicolored Shortest Paths in Graphs with Applications to Network Overlay Design Hongsik Choi and Hyeong-Ah Choi Department of Electrical Engineering and Computer Science George Washington University Washington,
More informationComputer Algorithms. NP-Complete Problems. CISC 4080 Yanjun Li
Computer Algorithms NP-Complete Problems NP-completeness The quest for efficient algorithms is about finding clever ways to bypass the process of exhaustive search, using clues from the input in order
More informationA Dynamic Programming Approach for Generating N-ary Reflected Gray Code List
A Dynamic Programming Approach for Generating N-ary Reflected Gray Code List Mehmet Kurt 1, Can Atilgan 2, Murat Ersen Berberler 3 1 Izmir University, Department of Mathematics and Computer Science, Izmir
More informationProtocols for Efficient Inference Communication
Protocols for Efficient Inference Communication Carl Andersen and Prithwish Basu Raytheon BBN Technologies Cambridge, MA canderse@bbncom pbasu@bbncom Basak Guler and Aylin Yener and Ebrahim Molavianjazi
More informationMALLET-Privacy Preserving Influencer Mining in Social Media Networks via Hypergraph
MALLET-Privacy Preserving Influencer Mining in Social Media Networks via Hypergraph Janani K 1, Narmatha S 2 Assistant Professor, Department of Computer Science and Engineering, Sri Shakthi Institute of
More informationLecture 15 An Arithmetic Circuit Lowerbound and Flows in Graphs
CSE599s: Extremal Combinatorics November 21, 2011 Lecture 15 An Arithmetic Circuit Lowerbound and Flows in Graphs Lecturer: Anup Rao 1 An Arithmetic Circuit Lower Bound An arithmetic circuit is just like
More informationConnected Identifying Codes for Sensor Network Monitoring
Connected Identifying Codes for Sensor Network Monitoring Niloofar Fazlollahi, David Starobinski and Ari Trachtenberg Dept. of Electrical and Computer Engineering Boston University, Boston, MA 02215 Email:
More informationInterconnection Network
Interconnection Network Recap: Generic Parallel Architecture A generic modern multiprocessor Network Mem Communication assist (CA) $ P Node: processor(s), memory system, plus communication assist Network
More informationThe positive minimum degree game on sparse graphs
The positive minimum degree game on sparse graphs József Balogh Department of Mathematical Sciences University of Illinois, USA jobal@math.uiuc.edu András Pluhár Department of Computer Science University
More informationSingle-Link Failure Detection in All-Optical Networks Using Monitoring Cycles and Paths
Single-Link Failure Detection in All-Optical Networks Using Monitoring Cycles and Paths Satyajeet S. Ahuja, Srinivasan Ramasubramanian, and Marwan Krunz Department of ECE, University of Arizona, Tucson,
More informationCS 598CSC: Combinatorial Optimization Lecture date: 2/4/2010
CS 598CSC: Combinatorial Optimization Lecture date: /4/010 Instructor: Chandra Chekuri Scribe: David Morrison Gomory-Hu Trees (The work in this section closely follows [3]) Let G = (V, E) be an undirected
More informationAdaptive Linear Programming Decoding
Adaptive Linear Programming Decoding Mohammad H. Taghavi and Paul H. Siegel ECE Department, University of California, San Diego Email: (mtaghavi, psiegel)@ucsd.edu ISIT 2006, Seattle, USA, July 9 14, 2006
More informationCOMBINATORIAL PROPERTIES OF THE HIGMAN-SIMS GRAPH. 1. Introduction
COMBINATORIAL PROPERTIES OF THE HIGMAN-SIMS GRAPH ZACHARY ABEL 1. Introduction In this survey we discuss properties of the Higman-Sims graph, which has 100 vertices, 1100 edges, and is 22 regular. In fact
More informationBuilding Alternate Multicasting Trees in MPLS Networks
Treasures at UT Dallas Eric Jonsson School of Engineering and Computer Science 5-2009 Building Alternate Multicasting Trees in MPLS Networks Limin Tang, et al. Follow this and additional works at: http://
More informationGraphical degree sequences and realizations
swap Graphical and realizations Péter L. Erdös Alfréd Rényi Institute of Mathematics Hungarian Academy of Sciences MAPCON 12 MPIPKS - Dresden, May 15, 2012 swap Graphical and realizations Péter L. Erdös
More informationA Catalogue of the Steiner Triple Systems of Order 19
A Catalogue of the Steiner Triple Systems of Order 19 Petteri Kaski 1, Patric R. J. Östergård 2, Olli Pottonen 2, and Lasse Kiviluoto 3 1 Helsinki Institute for Information Technology HIIT University of
More informationCompetitive Analysis of On line Randomized Call Control in Cellular Networks
Competitive Analysis of On line Randomized Call Control in Cellular Networks Ioannis Caragiannis Christos Kaklamanis Evi Papaioannou Abstract In this paper we address an important communication issue arising
More informationCOUNTING INDEPENDENT SETS IN SOME CLASSES OF (ALMOST) REGULAR GRAPHS
COUNTING INDEPENDENT SETS IN SOME CLASSES OF (ALMOST) REGULAR GRAPHS Alexander Burstein Department of Mathematics Howard University Washington, DC 259, USA aburstein@howard.edu Sergey Kitaev Mathematics
More informationScaling 10Gb/s Clustering at Wire-Speed
Scaling 10Gb/s Clustering at Wire-Speed InfiniBand offers cost-effective wire-speed scaling with deterministic performance Mellanox Technologies Inc. 2900 Stender Way, Santa Clara, CA 95054 Tel: 408-970-3400
More informationSocial Media Mining. Graph Essentials
Graph Essentials Graph Basics Measures Graph and Essentials Metrics 2 2 Nodes and Edges A network is a graph nodes, actors, or vertices (plural of vertex) Connections, edges or ties Edge Node Measures
More informationSHARP BOUNDS FOR THE SUM OF THE SQUARES OF THE DEGREES OF A GRAPH
31 Kragujevac J. Math. 25 (2003) 31 49. SHARP BOUNDS FOR THE SUM OF THE SQUARES OF THE DEGREES OF A GRAPH Kinkar Ch. Das Department of Mathematics, Indian Institute of Technology, Kharagpur 721302, W.B.,
More informationINTERNATIONAL JOURNAL OF COMPUTER ENGINEERING & TECHNOLOGY (IJCET)
INTERNATIONAL JOURNAL OF COMPUTER ENGINEERING & TECHNOLOGY (IJCET) International Journal of Computer Engineering and Technology (IJCET), ISSN 0976 6367(Print), ISSN 0976 6367(Print) ISSN 0976 6375(Online)
More informationClique coloring B 1 -EPG graphs
Clique coloring B 1 -EPG graphs Flavia Bonomo a,c, María Pía Mazzoleni b,c, and Maya Stein d a Departamento de Computación, FCEN-UBA, Buenos Aires, Argentina. b Departamento de Matemática, FCE-UNLP, La
More informationNear Optimal Solutions
Near Optimal Solutions Many important optimization problems are lacking efficient solutions. NP-Complete problems unlikely to have polynomial time solutions. Good heuristics important for such problems.
More informationAn Ecient Dynamic Load Balancing using the Dimension Exchange. Ju-wook Jang. of balancing load among processors, most of the realworld
An Ecient Dynamic Load Balancing using the Dimension Exchange Method for Balancing of Quantized Loads on Hypercube Multiprocessors * Hwakyung Rim Dept. of Computer Science Seoul Korea 11-74 ackyung@arqlab1.sogang.ac.kr
More informationDetermination of the normalization level of database schemas through equivalence classes of attributes
Computer Science Journal of Moldova, vol.17, no.2(50), 2009 Determination of the normalization level of database schemas through equivalence classes of attributes Cotelea Vitalie Abstract In this paper,
More informationComponents: Interconnect Page 1 of 18
Components: Interconnect Page 1 of 18 PE to PE interconnect: The most expensive supercomputer component Possible implementations: FULL INTERCONNECTION: The ideal Usually not attainable Each PE has a direct
More informationTenacity and rupture degree of permutation graphs of complete bipartite graphs
Tenacity and rupture degree of permutation graphs of complete bipartite graphs Fengwei Li, Qingfang Ye and Xueliang Li Department of mathematics, Shaoxing University, Shaoxing Zhejiang 312000, P.R. China
More informationDiversity Coloring for Distributed Data Storage in Networks 1
Diversity Coloring for Distributed Data Storage in Networks 1 Anxiao (Andrew) Jiang and Jehoshua Bruck California Institute of Technology Pasadena, CA 9115, U.S.A. {jax, bruck}@paradise.caltech.edu Abstract
More informationExtremal Wiener Index of Trees with All Degrees Odd
MATCH Communications in Mathematical and in Computer Chemistry MATCH Commun. Math. Comput. Chem. 70 (2013) 287-292 ISSN 0340-6253 Extremal Wiener Index of Trees with All Degrees Odd Hong Lin School of
More informationOn Integer Additive Set-Indexers of Graphs
On Integer Additive Set-Indexers of Graphs arxiv:1312.7672v4 [math.co] 2 Mar 2014 N K Sudev and K A Germina Abstract A set-indexer of a graph G is an injective set-valued function f : V (G) 2 X such that
More informationA REMARK ON ALMOST MOORE DIGRAPHS OF DEGREE THREE. 1. Introduction and Preliminaries
Acta Math. Univ. Comenianae Vol. LXVI, 2(1997), pp. 285 291 285 A REMARK ON ALMOST MOORE DIGRAPHS OF DEGREE THREE E. T. BASKORO, M. MILLER and J. ŠIRÁŇ Abstract. It is well known that Moore digraphs do
More informationDistributed Dynamic Load Balancing for Iterative-Stencil Applications
Distributed Dynamic Load Balancing for Iterative-Stencil Applications G. Dethier 1, P. Marchot 2 and P.A. de Marneffe 1 1 EECS Department, University of Liege, Belgium 2 Chemical Engineering Department,
More informationFactors to Consider When Designing a Network
Quality of Service Routing for Supporting Multimedia Applications Zheng Wang and Jon Crowcroft Department of Computer Science, University College London Gower Street, London WC1E 6BT, United Kingdom ABSTRACT
More informationFault-Tolerant Routing Algorithm for BSN-Hypercube Using Unsafety Vectors
Journal of omputational Information Systems 7:2 (2011) 623-630 Available at http://www.jofcis.com Fault-Tolerant Routing Algorithm for BSN-Hypercube Using Unsafety Vectors Wenhong WEI 1,, Yong LI 2 1 School
More informationAn Improved ACO Algorithm for Multicast Routing
An Improved ACO Algorithm for Multicast Routing Ziqiang Wang and Dexian Zhang School of Information Science and Engineering, Henan University of Technology, Zheng Zhou 450052,China wzqagent@xinhuanet.com
More informationThe Union-Find Problem Kruskal s algorithm for finding an MST presented us with a problem in data-structure design. As we looked at each edge,
The Union-Find Problem Kruskal s algorithm for finding an MST presented us with a problem in data-structure design. As we looked at each edge, cheapest first, we had to determine whether its two endpoints
More informationInternational Journal of Information Technology, Modeling and Computing (IJITMC) Vol.1, No.3,August 2013
FACTORING CRYPTOSYSTEM MODULI WHEN THE CO-FACTORS DIFFERENCE IS BOUNDED Omar Akchiche 1 and Omar Khadir 2 1,2 Laboratory of Mathematics, Cryptography and Mechanics, Fstm, University of Hassan II Mohammedia-Casablanca,
More informationChapter 4 Multi-Stage Interconnection Networks The general concept of the multi-stage interconnection network, together with its routing properties, have been used in the preceding chapter to describe
More informationComplexity Theory. IE 661: Scheduling Theory Fall 2003 Satyaki Ghosh Dastidar
Complexity Theory IE 661: Scheduling Theory Fall 2003 Satyaki Ghosh Dastidar Outline Goals Computation of Problems Concepts and Definitions Complexity Classes and Problems Polynomial Time Reductions Examples
More informationEnergy-efficient communication in multi-interface wireless networks
Energy-efficient communication in multi-interface wireless networks Stavros Athanassopoulos, Ioannis Caragiannis, Christos Kaklamanis, and Evi Papaioannou Research Academic Computer Technology Institute
More informationMulti-layer MPLS Network Design: the Impact of Statistical Multiplexing
Multi-layer MPLS Network Design: the Impact of Statistical Multiplexing Pietro Belotti, Antonio Capone, Giuliana Carello, Federico Malucelli Tepper School of Business, Carnegie Mellon University, Pittsburgh
More informationXiaoqiao Meng, Vasileios Pappas, Li Zhang IBM T.J. Watson Research Center Presented by: Payman Khani
Improving the Scalability of Data Center Networks with Traffic-aware Virtual Machine Placement Xiaoqiao Meng, Vasileios Pappas, Li Zhang IBM T.J. Watson Research Center Presented by: Payman Khani Overview:
More informationDECENTRALIZED LOAD BALANCING IN HETEROGENEOUS SYSTEMS USING DIFFUSION APPROACH
DECENTRALIZED LOAD BALANCING IN HETEROGENEOUS SYSTEMS USING DIFFUSION APPROACH P.Neelakantan Department of Computer Science & Engineering, SVCET, Chittoor pneelakantan@rediffmail.com ABSTRACT The grid
More informationLarge induced subgraphs with all degrees odd
Large induced subgraphs with all degrees odd A.D. Scott Department of Pure Mathematics and Mathematical Statistics, University of Cambridge, England Abstract: We prove that every connected graph of order
More informationRandom graphs with a given degree sequence
Sourav Chatterjee (NYU) Persi Diaconis (Stanford) Allan Sly (Microsoft) Let G be an undirected simple graph on n vertices. Let d 1,..., d n be the degrees of the vertices of G arranged in descending order.
More informationExpander Graph based Key Distribution Mechanisms in Wireless Sensor Networks
Expander Graph based Key Distribution Mechanisms in Wireless Sensor Networks Seyit Ahmet Çamtepe Computer Science Department Rensselaer Polytechnic Institute Troy, New York 12180 Email: camtes@cs.rpi.edu
More informationLoad Balancing Algorithm Based on Services
Journal of Information & Computational Science 10:11 (2013) 3305 3312 July 20, 2013 Available at http://www.joics.com Load Balancing Algorithm Based on Services Yufang Zhang a, Qinlei Wei a,, Ying Zhao
More informationInterconnection Network Design
Interconnection Network Design Vida Vukašinović 1 Introduction Parallel computer networks are interesting topic, but they are also difficult to understand in an overall sense. The topological structure
More informationApproximation Algorithms
Approximation Algorithms or: How I Learned to Stop Worrying and Deal with NP-Completeness Ong Jit Sheng, Jonathan (A0073924B) March, 2012 Overview Key Results (I) General techniques: Greedy algorithms
More informationA Numerical Study on the Wiretap Network with a Simple Network Topology
A Numerical Study on the Wiretap Network with a Simple Network Topology Fan Cheng and Vincent Tan Department of Electrical and Computer Engineering National University of Singapore Mathematical Tools of
More informationCost Model: Work, Span and Parallelism. 1 The RAM model for sequential computation:
CSE341T 08/31/2015 Lecture 3 Cost Model: Work, Span and Parallelism In this lecture, we will look at how one analyze a parallel program written using Cilk Plus. When we analyze the cost of an algorithm
More informationTypes of Degrees in Bipolar Fuzzy Graphs
pplied Mathematical Sciences, Vol. 7, 2013, no. 98, 4857-4866 HIKRI Ltd, www.m-hikari.com http://dx.doi.org/10.12988/ams.2013.37389 Types of Degrees in Bipolar Fuzzy Graphs Basheer hamed Mohideen Department
More informationTHE NUMBER OF REPRESENTATIONS OF n OF THE FORM n = x 2 2 y, x > 0, y 0
THE NUMBER OF REPRESENTATIONS OF n OF THE FORM n = x 2 2 y, x > 0, y 0 RICHARD J. MATHAR Abstract. We count solutions to the Ramanujan-Nagell equation 2 y +n = x 2 for fixed positive n. The computational
More informationLecture 23: Interconnection Networks. Topics: communication latency, centralized and decentralized switches (Appendix E)
Lecture 23: Interconnection Networks Topics: communication latency, centralized and decentralized switches (Appendix E) 1 Topologies Internet topologies are not very regular they grew incrementally Supercomputers
More informationLoad Balanced Optical-Network-Unit (ONU) Placement Algorithm in Wireless-Optical Broadband Access Networks
Load Balanced Optical-Network-Unit (ONU Placement Algorithm in Wireless-Optical Broadband Access Networks Bing Li, Yejun Liu, and Lei Guo Abstract With the broadband services increasing, such as video
More information3 Fat tree construction
Level-wise Scheduling Algorithm for Fat Tree Interconnection Networks Zhu Ding Union Switch & Signal Raymond R. Hoare Concurrent EDA, LLC Rami Melhem University of Pittsburgh Alex K. Jones University of
More informationScheduling Home Health Care with Separating Benders Cuts in Decision Diagrams
Scheduling Home Health Care with Separating Benders Cuts in Decision Diagrams André Ciré University of Toronto John Hooker Carnegie Mellon University INFORMS 2014 Home Health Care Home health care delivery
More informationA SOCIAL NETWORK ANALYSIS APPROACH TO ANALYZE ROAD NETWORKS INTRODUCTION
A SOCIAL NETWORK ANALYSIS APPROACH TO ANALYZE ROAD NETWORKS Kyoungjin Park Alper Yilmaz Photogrammetric and Computer Vision Lab Ohio State University park.764@osu.edu yilmaz.15@osu.edu ABSTRACT Depending
More informationMATHEMATICAL ENGINEERING TECHNICAL REPORTS. An Improved Approximation Algorithm for the Traveling Tournament Problem
MATHEMATICAL ENGINEERING TECHNICAL REPORTS An Improved Approximation Algorithm for the Traveling Tournament Problem Daisuke YAMAGUCHI, Shinji IMAHORI, Ryuhei MIYASHIRO, Tomomi MATSUI METR 2009 42 September
More information