Dynamic adaptation of response-time models for QoS management in autonomic systems
Introduction
Autonomic computing was introduced to decrease the difficulty of managing today's computing systems by setting high-level goals and letting the system manage itself to accomplish them (Ganek and Corbi, 2003, Kephart and Chess, 2003). These goals can be related to properties such as configuration, security or performance, and can be expressed in a SLA (Service Level Agreement) signed between two parties (for instance, the provider of a transactional system and its clients), to satisfy the clients’ demands from the system (Sahai et al., 2002). An SLA is composed of several Service Level Objectives (SLOs). A common SLO related to performance in transactional systems is an objective for the response time; specifically, a common SLO determines the 90-percentile of the response time for the operation of the system, that is, it determines the maximum response time for 90% of the system responses. Usually, the system has different functionalities or services with different requirements for their response times, which result in a different SLO for each kind of response.
If there are clients with different priorities, it is possible to improve the fulfilment rate of the SLOs for the most important clients by controlling the admission of lower priority clients. In order to automatically carry out this, an admission control subsystem must be inserted in the system (see Fig. 1). This subsystem must know the SLOs and the performance model of the system, so that when a new session is initiated by a client, the subsystem can decide whether to admit it in order to improve the fulfilling rate of the SLOs of high-priority clients. This rate can be predicted using the performance model.
For the system to work autonomically, changes in the system and in the workload characteristics must be incorporated to the performance model automatically, without the need for manual interventions. This means that the performance model must be automatically and dynamically adapted to reflect the new situation after the system or the workload characteristics change. In this work, we present a technique to carry out this dynamic adaptation.
Dynamic adaptation has several requirements. First, changes in the system must be detected; this means identifying when a permanent change (as opposed to a temporary change), has occurred. In addition, during transient states the system must continue to work stably; clients should not suffer the effects of erroneous decisions, such as a high number of rejections due to the admission control algorithm. Furthermore, the adaptation must be carried out in real-time, so it must not require excessive resource consumption. Finally, the dynamic adaptation technique must be easy to implement and to integrate in working systems. The technique presented in this paper meets all of these requirements.
The rest of the paper is organized as follows. In Section 2, related work is discussed. Section 3 describes the architecture of the target system. In Section 4, the dynamic adaptation technique proposed is presented, and its performance evaluation results are shown in Section 5. In Section 6, the conclusions summarize the work and present a discussion on its advantages and limitations.
Section snippets
Related work
QoS techniques need to know the state of the system in order to take control decisions. Using exclusively instantaneous measurements of the system resource usage is insufficient due to their extreme variability in Internet-based servers (Andreolini et al., 2008). In addition, in order to make effective decisions, such as whether to admit a new session, QoS techniques need to predict the state of the system depending on the decision made. Hence, modern QoS techniques usually rely on the
System architecture
The proposed technique for model adaptation is part of a strategy aimed at autonomically managing the QoS of a transactional system by means of admission control. The transactional system offers services of different kinds. Sessions encompass several transactions that the clients initiate to use the services offered by the system.
The QoS is defined by SLOs for each service. An SLO for a service states the expected maximum response time for 90% of the transactions for that service. Clients are
Dynamic adaptation strategy
The dynamic adaptation strategy implemented by the self-adjustment component must change the response-time model to reflect any changes introduced in the system or the workload characteristics. This can be carried out by measuring the response time of each transaction and adding the information obtained to the model. However, this creates several problems.
The first problem is that, if the model is updated continuously, a great amount of overhead will be added to the system due to the execution
Experimental results
In order to validate the technique proposed in this paper, a set of experiments has been carried out. The platform for the experiments implements the TPC-App benchmark (TPC, 2008), which simulates the activities of a business-to-business transactional application server that processes orders and retrieves information about product items. This benchmark has been chosen because it aims to represent common features of modern transactional systems.
The system has two main components (see Fig. 7): an
Discussion and conclusions
This paper has presented a technique for dynamic adaptation of response-time models used in autonomic computing systems. The technique is based on a Non-Negative Least Square (NNLS) regression model that is updated periodically with on-line measurements. The technique is part of an autonomic strategy that can be used for maintaining the SLOs of a system where the SLOs are expressed in terms of percentiles of the response time. The problems that arise when the information measured is used to
Acknowledgment
This research has been partially supported by the project TIN2008-06045-C02-01 of the Spanish National Plan for Research, Development and Innovation.
Joaquín Entrialgo is an Associate Professor in the Department of Informatics at the University of Oviedo, Spain. He received the MS degree in Computer Science in 1998 and the PhD degree in 2006. His main research interests are monitoring of real-time systems and performance evaluation of computer systems.
References (24)
- et al.
Models and framework for supporting runtime decisions in Web-based systems
ACM Trans. Web
(2008) - et al.
Nonnegativity Constraints in Numerical Analysis
(2007) - et al.
Quality of service management using generic modeling and monitoring techniques
Distrib. Syst. Eng.
(1997) - et al.
The dawning of the autonomic computing era
IBM Syst. J.
(2003) - et al.
A simulation model to develop QoS control strategies for application servers
Simulation
(2008) - et al.
A QoS control mechanism to provide service differentiation and overload protection to Internet scalable servers
IEEE Trans. Serv. Comput.
(2009) - et al.
Self-adjustment strategy for models used in autonomic transactional systems
- et al.
Modeling autonomic QoS control for grid service using Petri nets
- et al.
Performance characterization of black boxes with self-controlled load injection for simulation-based sizing
- et al.
Introducing queuing network-based performance awareness in autonomic systems
Yaksha: a self-tuning controller for managing the performance of 3-tiered web sites
The vision of autonomic computing
Computer
Cited by (7)
Extending OCCI for autonomic management in the cloud
2016, Journal of Systems and SoftwareCitation Excerpt :However, this coupling remains critical and challenging since Cloud environments “are composed of thousands of physical hosts and virtual machines, connected by many network elements” (Buyya et al., 2012). Autonomic Computing can reduce the difficulty of managing these environments (Entrialgo et al., 2011). Autonomic Computing (Jacob et al., 2004) is the ability to manage computing resources automatically and dynamically to respond to the requirements of the business based on SLA.
Optimal assignment of autonomic managers to cloud resources
2015, 10th IEEE Int. Conf. on Service Operations and Logistics, and Informatics, SOLI 2015 - In conjunction with ICT4ALL 2015Automatic generation of platforms in cloud computing
2012, Proceedings of the 2012 IEEE Network Operations and Management Symposium, NOMS 2012Quality of service for database in the cloud
2012, CLOSER 2012 - Proceedings of the 2nd International Conference on Cloud Computing and Services Science
Joaquín Entrialgo is an Associate Professor in the Department of Informatics at the University of Oviedo, Spain. He received the MS degree in Computer Science in 1998 and the PhD degree in 2006. His main research interests are monitoring of real-time systems and performance evaluation of computer systems.
Daniel F. García is a Professor in the Department of Informatics at the University of Oviedo, Spain. Since 1994 he has been leading the computer engineering area at the University of Oviedo. His current research interest is in the area of computer performance engineering and quality of service of computer systems. For the last ten years, Dr. Garcia has been conducting research projects in the area of information technologies at national and European levels. He is a member of the IEEE Computer Society.
Javier García is an Associate Professor in the Department of Informatics at the University of Oviedo, Spain, where he teaches courses in computer architecture. He received a PhD in electrical engineering from University of Oviedo. His research interests are in the performance evaluation of computer systems, and autonomic computing.
Manuel García received the Industrial Engineering degree and PhD degree with a thesis devoted to the analysis and modeling of HFC networks from the University of Oviedo, Spain, in 1992 and 2003, respectively. From 1993 to 1994 he worked at the university with a grant. In 1994, he became an Assistant Professor and, in 2001, he became an Associate Professor, in the Department of Informatics at the University of Oviedo. He has taken part in research projects on real-time inspection systems based on vision techniques, and performance analysis of HFC networks. His current research interests are computer and telecommunications systems performance evaluation.
Pablo Valledor obtained his MS degree in Computer Engineering in 2006 in the Polytechnic Engineering School of the University of Oviedo. He worked for the R&D department of CTIC Foundation (Centre for the Development of Information and Communication Technologies in Asturias) until February 2007, when he joined the ArcelorMittal R&D Technological Centre (CDT) as a research engineer. He also collaborates with the computer engineering area at the University of Oviedo. He has participated in several projects using his expertise in topics like data mining, optimization, semantic Web technologies, quality of service assurance in cluster environments and radioeletric signal simulation.
Professor Mohammad S. Obaidat (Fellow of IEEE and Fellow of SCS) is an internationally well known academic/researcher/ scientist. He received his PhD and MS degrees in computer engineering with a minor in computer science from The Ohio State University, Columbus, Ohio, USA. Dr. Obaidat is currently a Professor of computer science at Monmouth University, NJ, USA. His previous positions are Chair of the Department of computer science and Director of the graduate program at Monmouth University and a faculty member at the City University of New York. He has received extensive research funding and has published ten books and over four hundred and seventy five refereed technical articles in scholarly international journals and proceedings of international conferences, and currently working on three more books. He is the author of a new upcoming book: Wireless Sensor Networks (Cambridge University Press). He is also the editor of two new upcoming books: Cooperative Networking (John Wiley & Sons 2010) and pervasive computing and networking (John Wiley &Sons 2010). Prof. Obaidat is the author of the book entitled: “Fundamentals of Performance Evaluation of Computer and Telecommunications Systems,” by John Wiley & Sons in 2010. Dr. Obaidat is the Editor of the book entitled, “E-business and Telecommunication Networks”, published by Springer in 2008. He is the co-author of the book entitled, “Security of e-Systems and Computer Networks” published by Cambridge University Press in 2007. He is the co-author of the best selling book, “Wireless Networks” and “Multiwavelength Optical LANs” published by John Wily & Sons (2003). Obaidat is the editor of the book, APPLIED SYSTEM SIMULATION: Methodologies and Applications, published by Kluwer (now Springer) in 2003. Professor Obaidat has served as a consultant for several corporations and organizations worldwide. Mohammad is the Editor-in-Chief of the Wiley International Journal of Communication Systems and the FTRA Journal of Convergence. He served as an Editor of IEEE Wireless Communications from 2007–2010. Between 1991 and 2006, he served as a Technical Editor and an Area Editor of Simulation: Transactions of the Society for Modeling and Simulations (SCS) International, TSCS. He also served the Editorial Advisory Board of Simulation. He is now an Editor of the Wiley Security and Communication Networks Journal, Journal of Networks, International Journal of Information Technology, Communications and Convergence, IJITCC, Inderscience. He served on the International Advisory Board of the International Journal of Wireless Networks and Broadband Technologies, IGI-global. Prof. Obaidat is an Associate Editor/ Editorial Board member of seven other refereed scholarly journals including two IEEE Transactions, Elsevier Computer Communications Journal, Kluwer Journal of Supercomputing, SCS Journal of Defense Modeling and Simulation, Elsevier Journal of Computers and EE, International Journal of Communication Networks and Distributed Systems, The Academy Journal of Communications, International Journal of BioSciences and Technology and International Journal of Information Technology. He has served as a guest editor in numerous special issues of scholarly journals such as IEEE Transactions on Systems, Man and Cybernetics, SMC, IEEE Wireless Communications, IEEE Systems Journal, SIMULATION: Transactions of SCS, Elsevier Computer Communications Journal, Journal of C & EE, Wiley Security and Communication Networks, Journal of Networks, and International Journal of Communication Systems. Obaidat has served as the Steering Committee Chair, Advisory Committee Chair and Program Chair for numerous international conferences including the IEEE Int’l Conference on Electronics, Circuits and Systems, IEEE International Phoenix Conference on Computers and Communications, IEEE Int’l Performance, Computing and Communications Conference, IEEE International Conference on Computer Communications and Networks, SCS Summer Computer Simulation Conference, SCSC’97, SCSC98-SCSC2005, SCSC2006, the International Symposium on Performance Evaluation of Computer and Telecommunication Systems since its inception in 1998, International Conference on Parallel Processing, Honorary General Chair of the 2006 IEEE Intl. Joint Conference on E-Business and Telecommunications, ICETE2006. He has served as General Co-Chair of ICETE 2007-ICETE 2010. He has served as the Program Chair for the International Conference on Wireless Information Networks and Systems from 2008–present. He is the Co-Founder and Program Co-Chair of the International Conference on Data Communication Networking, DCNET since its inception in 2009. Obaidat has served as the General Chair of the 2007 IEEE International Conference on Computer Systems and Applications, AICCSA2007, the IEEE AICCSA 2009 Conference and the 2006 International Symposium on Ad hoc and Ubiquitous Computing (ISAHUC’06). He is the founder of the International Symposium on Performance Evaluation of Computer and Telecommunication Systems, SPECTS and has served as the General Chair of SPECTS since its inception. Obaidat has received a recognition certificate from IEEE. Between 1994 and 1997, Obaidat has served as distinguished speaker/visitor of IEEE Computer Society. Since, 1995 he has been serving as an ACM distinguished Lecturer and SCS distinguished Lecturer. Between 1996 and 1999, Dr. Obaidat served as an IEEE/ACM program evaluator of the Computing Sciences Accreditation Board/Commission, CSAB/CSAC. Obaidat is the Founder and first Chairman of SCS Technical Chapter (Committee) on PECTS (Performance Evaluation of Computer and Telecommunication Systems). He has served as the Scientific Advisor for the World Bank/UN Digital Inclusion Workshop—The Role of Information and Communication Technology in Development. Between 1995 and 2002, he has served as a member of the board of directors of the Society for Computer Simulation International. Between 2002 and 2004, he has served as Vice President of Conferences of the Society for Modeling and Simulation International SCS. Between 2004 and 2006, Prof. Obaidat has served as Vice President of Membership of the Society for Modeling and Simulation International SCS. Between 2006 and 2009, he has served as the Senior Vice President of SCS. Currently, he is the President of SCS. One of his recent co-authored papers has received the best paper award in the IEEE AICCSA 2009 international conference. He also received the best paper award for one of his papers accepted in IEEE GLOBCOM 2009 conference. Dr. Obaidat recently received the Society for Modeling and Simulation Intentional (SCS) prestigious McLeod Founder's Award in recognition of his outstanding technical and professional contributions to modeling and simulation.
He has been invited to give lectures and keynote speeches worldwide. His research interests are: Wireless Communications and Networks, Telecommunications and Networking systems, Security of Network, Information and Computer Systems, Security of e-based Systems, Performance Evaluation of Computer Systems, Algorithms and Networks, High Performance and Parallel Computing/Computers, Applied Neural Networks and Pattern Recognition, Adaptive Learning and Speech Processing. Recently, Prof. Obaidat has been awarded with Nokia Research Fellowship and the distinguished Fulbright Scholar Award. During 2004/2005, he was on sabbatical leave as Fulbright Distinguished Professor and Advisor to the President of Philadelphia University in Jordan, Dr. Adnan Badran. He latter became the Prime Minister of Jordan in April 2005 and served earlier as Vice President of UNESCO. Prof. Obaidat is the Fellow of the Society for Modeling and Simulation International SCS, and a Fellow of the Institute of Electrical and Electronics Engineers (IEEE). For more info, see: http://bluehawk.monmouth.edu/mobaidat/.
- 1
Fellow of IEEE and Fellow of SCS.