Available online at www.sciencedirect.com ScienceDirect

IERI Procedia 4 ( 2013 ) 261 – 267

2013 International Conference on Electronic Engineering and Architecture of a Novel Cloud Computing System Digital Or- Ganism System

Jianying Chen

College of Computer Science & Technology, Southwest University for Nationalities, Chengdu, China [email protected]

Abstract

To satisfy more and more demands of today and future applications, a novel cloud computing system, which is named Digital System (DOS), was developed successfully. With its own and management system, DOS provides distributed and parallel input/output, fault restoration, large throughput, high reliability/scalability, automatic management and other perfect performance. This paper introduces the novel system architecture and mainly focuses on the number of its servers, stations, clients and replicas.

© 20132013. The Published Authors. Published by Elsevier by Elsevier B.V. B.V. Selection and and peer peer review review under responsibilityunder responsibility of Information of EngineeringInformation Research Engineering Institute Research Institute

Keywords˖Cloud computing; digital organism system; architecture

1. Introduction

To satisfy the demands of today and future applications, cloud computing system must achieve high secu- rity, reliability and scalability, as well as intellectualized management, no special requirements about the in- frastructure (computers can be any PCs, PC servers, minicomputers or supercomputers), large enough throughput and storage capacity[1-4], etc. Thus, if the software of cloud computing system has organism like living things as follows, all can be solved successfully: a) Get information from the environment and then abstract it into knowledge; b) Self-adaptability[5], self-learning[6], self-dissemination and self-duplication; c) Simulate the gene and group of living things to automatically form ; d) Simulate the anti-viruses immune mechanisms of human being’s to realize self-anti-ruin architecture; e) Simulate the acquired genetic and ability of living things to make the system more “intelli-

2212-6678 © 2013 The Authors. Published by Elsevier B.V. Selection and peer review under responsibility of Information Engineering Research Institute doi: 10.1016/j.ieri.2013.11.037 262 Jianying Chen / IERI Procedia 4 ( 2013 ) 261 – 267

gent” and “healthy”; f) Simulate the gene nature of living things and continue its work when several units (not all) meet at- tacks or break down; g) Simulate the reproductive ability of living things to properly copy or reproduce replicas when it needs. h) Adopt the semantics to help people achieve logic analysis and decision-making. After years of study, a novel cloud computing system software named Digital Organism System (DOS) has successfully developed. DOS can fully meet all application requirements mentioned before. With the devel- opment of the theory and technology on science and its application, the nature of DOS will be further devel- oped. Next, we will discuss several key points about architecture of DOS. All problems seem to be simple but they are challenges for us and must be solved at first.

2. System architecture

All computers of DOS are connected by broadband network and installed with Digital Organism Operating System (DOOS), Digital Organism Database System (DODBS) and other correlative application software developed by us. All the servers placed nearby or divided by district can form one group, which is called sta- tion here. The number of servers within the station is determined by the requirements of application and bandwidth, which can vary from several to hundreds even boundless number. The scope of the station can also vary according to different requirements and applications. Take Chengdu for example, it may include the sub- urbs and the city proper, or just the city proper, or even one residential district. DOS topology is shown in figure 1. Here, we named computers or other equipments as client only because they enjoy services from servers. However, it is not the same as traditional C/S. The client defined here may be a PC, portable computer, digital set-top box and TV, or the TV that has in-built digital set-top box, etc. The set-top box mentioned here has the following characteristics: first, by this kind of set-top box, users can surf on the Internet, telephone by audio or video, and even choose real-time demands on request, multicast or broadcast TV programs by channels; secondly, if some of the servers break down, the services will not be stopped; thirdly, it can support many value-added services, such as speculating in stocks and shares, etc. One client can be intelligently and dynamically connected with several servers within a station or even with all servers of the whole station. In order to achieve high reliability and availability, it is never statically connected with one server. The system design guarantees that if one client cannot receive services from any servers formerly connected in the station then the service requests will be automatically and intelligently transferred to other stations, till the service is provided.

7HOHYLVLRQZLWK%R[ 7HOHYLVLRQ

67%

6WDWLRQ 3RUWDEOHFRPSXWHU 6WDWLRQ

%URDGEDQGQHWZRUN

6WDWLRQ 3& 6WDWLRQ 6WDWLRQ

Fig. 1 Architecture of DOS

The relation among stations is plane. All stations are equal and autonomous. Within one large enough Jianying Chen / IERI Procedia 4 ( 2013 ) 261 – 267 263 scope, e.g. the mainland of China, there must be many stations, but for one particular client request, the ser- vice should try to be worked locally. If the requirements can not be meet, the service will be transferred and provided by servers in other stations. If no servers can be found within the whole system, it is necessary to expand the system by same topology.

3. The number of servers within a station and the number of stations

Amount of effective concurrent video streaming and the concept of cycle will be used to discuss the num- ber of servers within a station and the number of stations in DOS.

1.1.The number of servers within one station

If there is enough bandwidth within one station, suppose the number of client being serviced by a server within the station is m, the of one client issuing a request is Pr, then the probability of servers to support i requests within one cycle is:

§m· iq ¨ ¸ i pp )1()( im ¨ ¸ r  r (1) © i ¹ The expectation of i requests being received and finished within one cycle is: § i  AA i · iE )( ¨ m m1 ¸m (2) ¨ i ¸ © Am ¹ Amount of effective concurrent video streaming is: ¦ iqiEB )()( (3) 0 dd mi The number of maximum concurrent video streaming one server supports within a cycle is n, the number of servers within a station is: nB (4) It has been verified that how large the number of concurrent video streaming a server supports is mainly determined by the I/O speed of hard disk. The relation between the number of servers within a station and the number of clients within a station is demonstrated in figure 2.

140 Pr=0.2 (a) n=100 140 Pr=0.2 (b) n=300 120 Pr=0.4 140 (c) n=1000 120 Pr=0.4 Pr=0.2 Pr=0.6 100 100 Pr=0.6 120 Pr=0.4 Pr=0.8 Pr=0.6 80 Pr=0.8 100 80 Pr=0.8 80 60 60 60 40 40 40 20 20

The number of servers 20 0 The number of servers 0 of servers The number 0 0 5000 10000 15000 0 12500 25000 37500 50000 0 50000 100000 150000 The number of clients The number of clients The number of clients

Fig. 2 The number of the servers and clients in a station 264 Jianying Chen / IERI Procedia 4 ( 2013 ) 261 – 267

1.2.The number of stations within the whole system

The number of stations within the whole system is mainly determined by the application requirements and the bandwidth among stations. The relation between any two stations in the system is plane, and there is broadband channel between them. Many kinds of applications can be developed on the basis of the platform of DOS, e.g. video-on-demand, real time TV broadcasting, audio telephone, video telephone, television-phone, searching information in In- ternet, electronic-office, electronic-commerce, bank operating etc. The author believes it is reasonable to take television-phone as an example to study the number of stations within the system. One television-phone call between two clients is like one television streaming. Let’s define the duration of one television-phone call as a cycle (take its average value). If the probability of one client making a televi- sion-phone call between stations is Pt, then within one cycle the probability of making j television-phone calls from one station is: m § m ·  j jq ¨ 2 ¸ j pp )1()( 2 ¨ ¸ t  t (5) © j ¹ In the formula, m/2 is considered as one call being made between two clients. Within one cycle the expec- tation value of j television-phone calls being finished is:  AA jj m m 1 22 m jE j )()( u (6) Am 2 2 Amount of effective concurrent television-phone is:

x ¦ jqjEB )()( (7) m 0 jdd 2 If the number of the stations is k, the bandwidth one television-phone occupies is d, then the total number of the maximum bandwidth required is: k ¦dBx (8) x 1 If the bandwidth between two stations is F, then there should be: k 2 k t ¦dBFC x (9) x 1 According to formula (9), given the broadband network environment (condition), the number of stations within the system can be got. Suppose the bandwidth of one television-phone call occupies is 1Mbps, the probability of one client mak- ing television-phone call between stations is 0.8, 0.6, 0.4, 0.2 respectively, then the relation of the number of stations within the system and the number of television-phone clients among stations can be demonstrated in figure 3. Jianying Chen / IERI Procedia 4 ( 2013 ) 261 – 267 265

(b) F=1Tbps (c) F=10Tbps 40 (a) F=100Gbps 140 Pt=0.2 80 Pt=0.2 35 Pt=0.2 120 Pt=0.4 70 Pt=0.4 60 30 Pt=0.4 100 Pt=0.6 Pt=0.6 50 25 Pt=0.6 Pt=0.8 Pt=0.8 Pt=0.8 80 40 20 60 30 15 40 20 10 10 5 20 The number of stations of number The The number ofstations 0 0 stations of The number 0 0 7500 15000 22500 0 30000 60000 90000 0 300000 600000 900000 The number of clients The number of clients The number of clients

Fig. 3 The relation of the number of stations within the system and the number of television-phone clients among stations

4. The number of copies and connections of client

One message (e.g. one file) must have its copy. The number of copy intelligently and dynamically changes or is determined by the requirement of anti-ruin capability or self-learning. All the copies of one message cannot be put at the same place. For example, in order to cope with the possibility of partial power-off, they cannot be put at the same power-supplying district; same as facing the war they can not be put in the same city etc. Definition 1: Anti-ruin Capability The survival abilities of DOS when encountering attacks. Suppose the system is composed of C units, at the time period ǻt1, the system is attacked by many attack- y ers, if the probability of hitting a unit is Pu, then the probability of y units being hit is Pu , the number of hits yPy yPy is yP y , its influence to the system is u , 1 u is the arithmetic expression of anti-ruin capability. u C C The relation between the system anti-ruin capability and the number of units destroyed is demonstrated in figure 4. 1.0 0.9 0.8 0.7 0.6 Pu = 0.2 0.5 Pu = 0.4 0.4 Pu = 0.6

Anti-ruin capability 0.3 Pu = 0.8 0.2 0246810 The number of units destroyed

Fig. 4 System anti-ruin capability and the number of the units destroyed For the system, the unit represents station; for the station, the unit represents server; for the file, the unit represents copy; for the client, the unit represents one connection of the client with different servers or sta- tions. For the availability, ȝ(rebuild rate)>Ȝ (breakdown rate); as for the anti-ruin capability, ȝ>Ȝ cannot be re- quired at the ǻt1 period because of parallel attacking. Definition 2: Attacking Interval It refers to the interval between the beginning of the first attacking and the beginning of the second attacking, and it is marked ǻt2. Within the time period of ǻt2, parallel restoration technology must be adopted to guarantee ȝ>Ȝ, or the sys- 266 Jianying Chen / IERI Procedia 4 ( 2013 ) 261 – 267

tem will lose survival ability. For example, if the copy number of a file is 3, one or two copies have been hit in one attacking. Then, one or two copies must be restored or reproduced at the ǻt2 period followed. y Just as animals have different sexes to maintain their survival, digital organism system has u t 2yPA y to do that. The system has the ability of teach oneself and self-adaptability, once it has found  yPA u is close to 2 or some emergent requirements, it should increase its reproduction. When there is too much volume, it automatically eliminates some, like what is happening in nature, as the animal system automatically main- tains balance according to the needs [7]. When it has found that the anti-ruin capability is too low, it should adopt measures such as increasing the number of units to improve the system anti-ruin capability [8]. Based on the anti-ruin capability expression, the number of copies and the connection number of clients can be determined. For example, if one thought the number of copies is 3, and the anti-ruin capability is low and will affect the survival of the system, then the number can be increased to 4.

5. Conclusion

To meet all demands of users and applications mentioned in Introduction of [9-13], we set about our re- search and development on DOS (Digital Organism System) since several years ago [14-15]. Compared with single computer, minicomputer, supercomputer, 1:1 protection system warm backup, network computing, grid computing and cluster system etc, DOS is a novel cloud computing system. It is com- posed of a series of software, including Digital Organism Operating System (DOOS), Digital Organism Database System (DODBS) and many application software based on it. It must be emphasized is that, DOS introduces social scientific methodology, simulates living things and its living groups (especially human being) to achieve organism characteristics, such as anti-viruses immune mechanisms, self-adaptability, self-learning, self-duplication and so on. This paper is only about the architecture of the system and mainly focuses on the number of its servers, stations, clients and replicas. Next, many ways, means and mechanisms adopted and contrived in DOS will be discussed from different angle. Though we have achieved several breakthroughs and successful applications at present, there is still a long way ahead.

References

[1] Wu, B. Dovey, M. Ng, etc., A Web/grid portal implementation of BioSimGrid: a biomolecular simulation database Information Technology: Coding and Computing, 2004. Proceedings. ITCC 2004. International Conference on Volume 2, 5-7 April 2004 Page(s): 50-54 Vol.2. [2] Young-Hyun Moon, Fast service restoration under deregulated environment, Transmission and Distribution Conference and Exhibition 2002: Asia Pacific. IEEE/PES, Volume: 1, 6-10 Oct. 2002 Pages: 574-577vol.1. [3] Borcea, C. Intanagonwiwat, C. Saxena, A. Iftode, L., Self-routing in pervasive computing environments using smart messages Pervasive Computing and Communications, 2003. (PerCom 2003). Proceedings of the First IEEE International Conference on 23-26 March 2003 Page(s): 87–96. [4] Lu Sheng-Min, YANG Xun-Nian, WANG Guo-Zhao, A Layered Surfaces Reconstruction from Large Vomuletric Datasets, Journel of Software, vol. 14, No. 8, 2003, 1448-1455. [5] Haiying Wang, Azuaje, F. Black, N., An integrative and interactive framework for improving biomedical pattern discovery and , Information Technology in Biomedicine, IEEE Transactions on, Volume 8, Issue 1, March 2004 Page(s):16 – 27. Jianying Chen / IERI Procedia 4 ( 2013 ) 261 – 267 267

[6] Yang Zhaosheng; Sun Jianping; Yang Chunxiao, Multi-agent urban expressway control system based on generalized knowledge-based model, Intelligent Transportation Systems, 2003. Proceedings. 2003 IEEE, Volume 2, 12-15 Oct. 2003 Page(s):1759 - 1763 vol.2. [7] Ray, T.S. Hart, J. Evolution of differentiated multi-threaded digital organisms. Intelligent Robots and Systems, 1999. IROS '99. Proceedings. 1999 IEEE/RSJ International Conference on. Volume 1, 17-21 Oct. 1999 Page(s):1 - 10 vol.1. [8] Ball, N.R.; Warwick, K.;. Using self-organising feature maps for the control of artificial organisms. and Applications, IEE Proceedings D [see also IEE Proceedings-Control Theory and Applications] , Volume: 140 , Issue: 3 , May 1993 Pages:176 – 180. [9] Dolev, D., Malki, D.; Yarom, Y.;. Warm backup using snooping. Distributed and Networked Environments, 1994. Proceedings., First International Workshop on Services in , 27-28 June 1994 Pages:60 – 65. [10] Maheshwaran, M., Ali, S., A of network computing systems. Computer , Volume: 37 , Issue: 10 , Oct. 2004 Pages:115 – 119. [11] Browne, J.C., Grid computing as applied distributed computation: a graduate seminar on Internet and Grid computing. Cluster Computing and the Grid, 2004. CCGrid 2004. IEEE International Symposium on , 19-22 April 2004 Pages:239-245. [12] Patnaik, L.M., High performance cluster computing. Concurrency, IEEE [see also IEEE Parallel & Distributed Technology] , Volume: 8 , Issue: 1 , Jan.-March 2000 Pages:86 – 87. [13] LIU Xinsong, The performance Research of the Distributed Parallel Server System with Distributed Parallel I/O Interface, ACTA ELECTRONICA SINICA, Vol. 30, No. 12, Dec. 2002, 1808-1810. [14] LIU Xinsong, Qiu Yuanjie, Yang Feng, etc., Architecture Research of No-Stop Computer System. Chinese Journal of Electronic,2004.Oct.pp.:559-564. [15] LIU Xin-song, WANG Dian-Yin, The Anti-ruin Architecture Research of the Digital Organism System, Dr & Specialist Forum of SiChuan, 2004, 330-334.