Skip to main content
SpringerLink
Log in

Neural Adaptive Control in Application Service Management Environment

  • Original Paper
  • Published:
Evolving Systems Aims and scope Submit manuscript

Abstract

We introduce a learning controller framework for adaptive control in application service management environments and explore its potential. Run-time metrics are collected by observing the enterprise system during its normal operation and load tests are persisted creating a knowledge base of real system states. Equipped with such knowledge the proposed framework associates system states and high/low service level agreement values with successful/unsuccessful control actions. These associations are used to induce decision rules, which help generating training sets for a neural networks-based control decision module that operates in the application run-time. Control actions are executed in the background of the current system state, which is then again monitored and stored extending the system state repository/knowledge base, and evaluating the correctness of the control actions frequently. This incremental learning leads to evolving controller behavior by taking into account consequences of earlier actions in a particular situation, or other similar situations. Our tests demonstrate that this controller is able to adapt to changing run-time conditions and workloads based on SLA definitions and is able to control the instrumented system under overloading effectively.

This is a preview of subscription content, log in via an institution to check access.

Access this article

Log in via an institution

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Institutional subscriptions

Fig. 1
Fig. 2
Fig. 3
Fig. 4
Fig. 5
Fig. 6
Fig. 7
Fig. 8
Fig. 9

Similar content being viewed by others

Notes

  1. The system resources can be modified in order to change important run-time characteristics—this scenario is not discussed in this paper.

  2. The term controller is used as a name of a function of the proposed framework. Due to distributed nature of the framework and the lack of a model of the enterprise system, the function is split into two software components and is deployed in isolation, see Sects. 5.4 and 5.6, where the evaluator, which generates control rules by learning from available data, and the controller API, which coordinates work of actuator agents, are described.

  3. Allmon (web, 2012a) is a Java monitoring tool, freely available on Apache License 2.0 (Apache 2004).

  4. AOP introduces a few useful concepts improving separability of the monitoring and the controller code from the system functionality. AOP bases on separation of cross-cutting concerns, applying advices, point-cut, and aspects (Kiczales et al. 1997; Laddad 2009).

  5. http://httpd.apache.org/docs/trunk/filter.html in Java Servlet specification since version 2.3 (Coward and Yoshida 2003).

  6. Java-ML is a Java Machine Learning library, available of GPL http://java-ml.sourceforge.net/content/feature-selection.

  7. 240 system state points in an hour generate 5.7 K in a day, 40 K in a week, 2.1 M in a year.

  8. Merge sets of states of high SLA values and “good” control to promote these control actions.

  9. Merge sets of states of low SLA values and “bad” control to prevent harmful control decisions.

  10. Trained networks are sent to the controller actuators, so they are directly available in the controlled application run- time.

  11. Select m system states S of highest SLAs for “good” control states, where penalties for the control SLA apc (S) were lower than total of SLAs in the state neighborhood.

  12. Select m states of lowest SLAs for “bad” control states, where penalties for the control SLA apc (S) were higher than SLAs (too harmful control)—these rules instruct to avoid applying any control in such states.

  13. A “good” control decision is the one that makes the situation better, so select system states S where total of SLAs is lower for similar state parameters.

  14. A “bad” control decision is the one that makes the situation worse, so select states where SLAs are higher for similar state parameters.

  15. In the simplest form both system states are removed. There have been more variants of resolving conflicts researched as well.

  16. Neuroph 2.4 is an open source Java neural network framework http://neuroph.sourceforge.net/. It contains run-time level reference to another NN library, called Encog http://code.google.com/p/encog-java/. Both of them are published under Apache 2.0 license Apache (2004).

  17. The ddd sample application (web, 2012b) was used as a model enterprise application with a few modifications mainly concerning the load characteristics. It is a sample application which was developed for demonstration of Domain-Driven Design concepts introduced by Evans (2004), and is freely available under the MIT License http://www.opensource.org/licenses/MIT.

  18. Grinder (Aston and Fitzgerald 2005) is a general purpose Java load testing framework specialized in using many load injector machines for flexible load tests (jython scripts) in distributed environments/systems—it is freely available under the BSD license, http://www.opensource.org/licenses/bsd-license.php.

  19. During the simulations the evaluator needed from 20 to 90 s to process the above described algorithm. When the evaluator was running CPU was significantly utilized, which was impacting the application running under load and effectively was changing whole system characteristics. The system had to adapt to such conditions.

References

  • (2012a) Allmon, a generic system collecting and storing metrics used for performance and availability monitoring. http://code.google.com/p/allmon/

  • (2012b) DDD sample application, the project is a joint effort by Eric Evans’ company domain language and the Swedish software consulting company Citerus. http://dddsample.sourceforge.net/

  • Abdelzaher T, Lu C (2000) Modeling and performance control of internet servers. In: Proceedings of the 39th IEEE conference on decision and control, vol 3. IEEE, pp 2234–2239

  • Abdelzaher T, Shin K, Bhatti N (2002) Performance guarantees for web server end-systems: a control-theoretical approach. IEEE Trans Parallel Distrib Syst 13(1):80–96

    Article  Google Scholar 

  • Abdelzaher T, Stankovic J, Lu C, Zhang R, Lu Y (2003) Feedback performance control in software services. IEEE Control Syst 23(3):74–90

    Article  Google Scholar 

  • Abeel T, Van de Peer Y, Saeys Y (2009) Java-ml: a machine learning library. J Mach Learn Res 10:931–934

    MathSciNet  MATH  Google Scholar 

  • Abrahao B, Almeida V, Almeida J, Zhang A, Beyer D, Safai F (2006) Self-adaptive sla-driven capacity management for internet services. In: Network operations and management symposium, 2006. NOMS 2006. 10th IEEE/IFIP, IEEE, pp 557–568

  • Apache (2004) Version 2.0. The Apache Software Foundation: Apache License, Version 20

  • Aston P, Fitzgerald C (2005) The grinder, a java load testing framework. http://grinder.sourceforge.net/

  • Åström KJ, Wittenmark B (2008) Adaptive control. Dover Publications

  • Bertoncini M, Pernici B, Salomie I, Wesner S (2011) Games: green active management of energy in it service centres. Information systems evolution, pp 238–252

  • Bigus J (1994) Applying neural networks to computer system performance tuning. In: 1994 IEEE international conference on neural networks, 1994. IEEE world congress on computational intelligence, vol 4. IEEE, pp 2442–2447

  • Bigus J, Hellerstein J, Jayram T, Squillante M (2000) Autotune: a generic agent for automated performance tuning. Practical application of intelligent agents and multi agent technology

  • Bodık P, Griffith R, Sutton C, Fox A, Jordan M, Patterson D (2009) Statistical machine learning makes automatic control practical for internet datacenters. In: Proceedings of the 2009 conference on hot topics in cloud computing, HotCloud, vol 9

  • Boniface M, Nasser B, Papay J, Phillips S, Servin A, Yang X, Zlatev Z, Gogouvitis S, Katsaros G, Konstanteli K, et al (2010) Platform-as-a-service architecture for real-time quality of service management in clouds. In: 2010 Fifth international conference on internet and web applications and services (ICIW). IEEE, pp 155–160

  • Brown B, Chui M, Manyika J (2011) Are you ready for the era of big data? McKinsey Global Institute

  • Buyya R, Yeo C, Venugopal S (2008) Market-oriented cloud computing: vision, hype, and reality for delivering it services as computing utilities. In: 10th IEEE international conference on high performance computing and communications, 2008. HPCC’08. IEEE, pp 5–13

  • Buyya R, Yeo C, Venugopal S, Broberg J, Brandic I (2009) Cloud computing and emerging it platforms: vision, hype, and reality for delivering computing as the 5th utility. Future Gen Comput Syst 25(6):599–616

    Article  Google Scholar 

  • Chen Y, Gmach D, Hyser C, Wang Z, Bash C, Hoover C, Singhal S (2010) Integrated management of application performance, power and cooling in data centers. In: 2010 IEEE network operations and management symposium (NOMS). IEEE, pp 615–622

  • Cleveland W, Devlin S (1988) Locally weighted regression: an approach to regression analysis by local fitting. J Am Stat Assoc 83(403):596–610

    Article  MATH  Google Scholar 

  • Coward D, Yoshida Y (2003) Java servlet specification version 2.3. Sun Microsystems

  • Emeakaroha V, Brandic I, Maurer M, Dustdar S (2010) Low level metrics to high level slas-lom2his framework: bridging the gap between monitored metrics and sla parameters in cloud environments. In: 2010 international conference on high performance computing and simulation (HPCS). IEEE, pp 48–54

  • Evans E (2004) Domain-driven design: tackling complexity in the heart of software. Addison-Wesley Professional

  • Fodor I (2002) A survey of dimension reduction techniques, vol 9. Center for Applied Scientific Computing, Lawrence Livermore National Laboratory, pp 1–18

  • Grinshpan L (2012) Solving enterprise applications performance puzzles: queuing models to the rescue. Wiley Online Library

  • Guyon I, Elisseeff A (2003) An introduction to variable and feature selection. J Mach Learn Res 3:1157–1182

    MATH  Google Scholar 

  • Haines S (2006) Pro Java EE 5 performance management and optimization. Apress

  • Hellerstein J (2004) Challenges in control engineering of computing systems. In: American Control Conference, 2004. Proceedings of the 2004, IEEE 3:1970–1979

  • Hellerstein J, Parekh S, Diao Y, Tilbury D (2004) Feedback control of computing systems. Wiley–IEEE Press

  • Herrnstein R (1970) On the law of effect. J Exp Anal Behav 13(2):243

    Article  Google Scholar 

  • Kandasamy N, Abdelwahed S, Hayes J (2004) Self-optimization in computer systems via on-line control: application to power management. In: Proceedings of the international conference on autonomic computing, 2004. IEEE, pp 54–61

  • Kiczales G, Lamping J, Mendhekar A, Maeda C, Lopes C, Loingtier J, Irwin J (1997) Aspect-oriented programming. ECOOP’97 object-oriented programming, pp 220–242

  • Kowall J, Cappelli W (2012) Magic quadrant for application performance monitoring. Gartner Research ID:G00232180

  • Kusic D, Kephart J, Hanson J, Kandasamy N, Jiang G (2009) Power and performance management of virtualized computing environments via lookahead control. Cluster Comput 12(1):1–15

    Article  Google Scholar 

  • Laddad R (2009) Aspectj in action: enterprise AOP with spring applications. Manning Publications Co.

  • Lu Y, Abdelzaher T, Lu C, Tao G (2002) An adaptive control framework for qos guarantees and its application to differentiated caching. In: Tenth IEEE international workshop on quality of service, 2002. IEEE, pp 23–32

  • Lu Y, Abdelzaher T, Lu C, Sha L, Liu X (2003) Feedback control with queueing-theoretic prediction for relative delay guarantees in web servers. In: Proceedings of the 9th IEEE real-time and embedded technology and applications symposium, 2003. IEEE, pp 208–217

  • Muggleton S (1999) Inductive logic programming: issues, results and the challenge of learning language in logic. Artif Intell 114(1):283–296

    Article  MATH  Google Scholar 

  • Muggleton S, De Raedt L (1994) Inductive logic programming: theory and methods. J Logic Program 19:629–679

    Article  MathSciNet  Google Scholar 

  • Parekh S, Gandhi N, Hellerstein J, Tilbury D, Jayram T, Bigus J (2002) Using control theory to achieve service level objectives in performance management. Real Time Syst 23(1):127–141

    Article  MATH  Google Scholar 

  • Park L, Baek J, Woon-Ki Hong J (2001) Management of service level agreements for multimedia internet service using a utility model. IEEE Commun Mag 39(5):100–106

    Article  Google Scholar 

  • Patel P, Ranabahu A, Sheth A (2009) Service level agreement in cloud computing. In: Cloud workshops at OOPSLA

  • Powley W, Martin P, Ogeer N, Tian W (2005) Autonomic buffer pool configuration in postgresql. In: 2005 IEEE international conference on systems, man and cybernetics, vol 1. IEEE, pp 53–58

  • Ripley BD (2012) Lowees scatter plot smoothing, r statistical data analysis, r documentation. http://stat.ethz.ch/R-manual/R-patched/library/stats/html/lowess.html

  • Saeys Y, Abeel T, Van de Peer Y (2008) Robust feature selection using ensemble feature selection techniques. Machine learning and knowledge discovery in databases, pp 313–325

  • Skinner B (1938) The behavior of organisms: an experimental analysis

  • Skinner B (1963) Operant behavior. Am Psychol 18(8):503

    Article  Google Scholar 

  • Slotine JJE, Li W et al (1991) Applied nonlinear control, vol 1. Prentice Hall, New Jersey

  • Stantchev V, Schröpfer C (2009) Negotiating and enforcing qos and slas in grid and cloud computing. Advances in grid and pervasive computing, pp 25–35

  • Sun D, Chang G, Li F, Wang C, Wang X (2011) Optimizing multi-dimensional qos cloud resource scheduling by immune clonal with preference. Acta Electron Sin 8:018

    Google Scholar 

  • Sutton R (1984) Temporal credit assignment in reinforcement learning. PhD thesis

  • Sutton R, Barto A (1998) Reinforcement learning: an introduction, vol 1. Cambridge University Press, Cambridge

  • Thorndike E, Bruce D (1911) Animal intelligence: experimental studies. Transaction Pub

  • Vaquero L, Rodero-Merino L, Caceres J, Lindner M (2008) A break in the clouds: towards a cloud definition. ACM SIGCOMM Comput Commun Rev 39(1):50–55

    Article  Google Scholar 

  • Wang Z, Chen Y, Gmach D, Singhal S, Watson B, Rivera W, Zhu X, Hyser C (2009) Appraise: application-level performance management in virtualized server environments. IEEE Trans Netw Serv Manag 6(4):240–254

    Article  Google Scholar 

  • Watkins C (1989) Learning from delayed rewards. PhD thesis, King’s College, Cambridge

  • Welsh M, Culler D (2002) Overload management as a fundamental service design primitive. In: Proceedings of the 10th workshop on ACM SIGOPS European workshop. ACM, pp 63–69

  • Welsh M, Culler D (2003) Adaptive overload control for busy internet servers. In: Proceedings of the 4th USENIX conference on internet technologies and systems, vol 2

  • Xiong P, Wang Z, Jung G, Pu C (2010) Study on performance management and application behavior in virtualized environment. In: Network operations and management symposium (NOMS), 2010 IEEE. IEEE, pp 841–844

  • Zhang A, Santos P, Beyer D, Tang H (2002a) Optimal server resource allocation using an open queueing network model of response time. HP laboratories Technical Report, HPL2002301

  • Zhang R, Lu C, Abdelzaher T, Stankovic J (2002b) Controlware: a middleware architecture for feedback control of software performance. In: Proceedings of the 22nd international conference on distributed computing systems, 2002. IEEE, pp 301–310

Download references

Acknowledgments

This work was partially sponsored by Solid Software Solutions (http://www.solidsoftware.pl/).

Author information

Authors and Affiliations

  1. Department of Computer Science and Information Systems, Birkbeck, University of London, Malet Street, London, WC1E 7HX, UK

    Tomasz D. Sikora & George D. Magoulas

Authors
  1. Tomasz D. Sikora
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. George D. Magoulas
    View author publications

    You can also search for this author in PubMed Google Scholar

Corresponding author

Correspondence to Tomasz D. Sikora.

Rights and permissions

Reprints and permissions

About this article

Cite this article

Sikora, T.D., Magoulas, G.D. Neural Adaptive Control in Application Service Management Environment. Evolving Systems 4, 267–287 (2013). https://doi.org/10.1007/s12530-013-9089-2

Download citation

  • Received:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1007/s12530-013-9089-2

Keywords

Search

Navigation