Application Performance Monitoring & Architecture Discovery with Kieker

Transcription

1 Application Performance Monitoring & Architecture Discovery with Kieker Prof. Dr. Wilhelm (Willi) Hasselbring Competence Cluster Software Systems Engineering Software Engineering Group, Kiel University WAIS Seminar Southampton, April, 3 rd 04 W. Hasselbring (CAU) Kieker Southampton / 5

2 Thanks to Contributors Several persons contributed to Kieker and the contents of this presentation in the past eight years: Tillmann (Till) Bielefeld, Peer Brauer, Philipp Döhring, Jens Ehlers, Nils Ehmke, Florian Fittkau, Thilo Focke, Sören Frey, Tom Frotscher, Henry Grow, André van Hoorn, Reiner Jung, Benjamin Kiel, Dennis Kieselhorst, Holger Knoche, Arnd Lange, Marius Löwe, Marco Lübcke, Felix Magedanz, Nina Marwede, Robert von Massow, Jasminka Matevska, Oliver Preikszas, Sönke Reimer, Bettual Richter, Matthias Rohr, Nils Sommer, Lena Stöver, Jan Waller, Robin Weiß, Björn Weißenfels, Matthias Westphal, Christian Wulf Alphabetic list of people who contributed in various forms and intensity W. Hasselbring (CAU) Kieker Southampton / 5

3 Agenda Dynamic Software Analysis Kieker Monitoring & Analysis Framework 3 Application Performance Monitoring 4 Reverse Engineering 5 Summary and Future Work W. Hasselbring (CAU) Kieker Southampton 3 / 5

4 Static Analysis of Software Systems To distinguish it from dynamic analysis Dynamic Software Analysis Static analysis may is applied to the program code to check the programs quality for various concerns, some examples: Clone detection Bauhaus Code metrics JavaNCSS Coding conventions Checkstyle Fault patterns FindBugs Architecture analysis Sotograph W. Hasselbring (CAU) Kieker Southampton 4 / 5

5 Dynamic Analysis of Software Systems "Dynamic Software Analysis Dynamic analysis, or the analysis of data gathered from a running program, has the potential to provide an accurate picture of a software because it exposes the system system's actual behavior. This picture can range from up to class-level details high-level architectural views [...]. " (Cornelissen et al. 009) W. Hasselbring (CAU) Kieker Southampton 5 / 5

6 Dynamic Analysis of Software Systems (cont d) "Dynamic Software Analysis Among the benefits over static analysis are the availability of runtime information and, in the context of object-oriented software, the exposure of object identities and the actual resolution of late binding. A drawback is that dynamic analysis can only provide a partial picture of the system, i.e., the results obtained are valid for the scenarios that were exercised during the analysis. " (Cornelissen et al. 009) W. Hasselbring (CAU) Kieker Southampton 6 / 5

7 Dynamic and Static Analysis of Software Dynamic Software Analysis Static analysis alone is not sufficient to study the internal behavior of software systems comprehensively Continuous monitoring allows to gather a system s actual runtime behavior resulting from production usage profiles So, how to collect runtime data from running systems? Instrumentation Profiling employed in development environments Monitoring employed in production environments An comprehension of the requirements of system operation is required to do the right things at development time Instrument upfront! Integrate with continuous integration and delivery. Have you heard of DevOps? W. Hasselbring (CAU) Kieker Southampton 7 / 5

8 How does Monitoring work? Example instrumentation: Measuring response times Dynamic Software Analysis Mean Median Instrumentation public static void searchbook() { 86 long tin = System.currentTimeMillis(); 87 Catalog.getBook(false); 88 long tout = System.currentTimeMillis(); 89 System.out.printf("Catalog;getBook;%s;%s", tin, tout); 90 } 9 Catalog;getBook; ; Catalog;getBook; ; Catalog;getBook; ; Catalog;getBook; ; Catalog;getBook; ; Program execution Visualization of results Monitoring log Statistical analysis Challenges Flexibility (instrumentation, changing log formats), runtime overhead, data volume, data distribution, online analysis, legacy systems,... W. Hasselbring (CAU) Kieker Southampton 8 / 5

9 Continuous Monitoring Design Decisions To cope with the previously listed challenges [van Hoorn et al. 009a] Dynamic Software Analysis Integration into the engineering process Model-Driven Instrumentation & Analysis. Type of Monitoring Probes Depends on analysis goals Number and Position of Monitoring Points Trade-off between performance overhead and information quality Consideration of the Induced Monitoring Overhead During continuous operation: only small overhead acceptable Physical Location of the Monitoring Log/Stream E.g., database, file system, queue, distributed deployment Intrusiveness of Instrumentation Don t mix monitoring logic with business functionality AOP W. Hasselbring (CAU) Kieker Southampton 9 / 5

10 Agenda Kieker Monitoring & Analysis Framework Dynamic Software Analysis Kieker Monitoring & Analysis Framework 3 Application Performance Monitoring 4 Reverse Engineering Reverse Engineering of Java EE Reverse Engineering of C# Reverse Engineering of Visual Basic 6 Reverse Engineering of COBOL Reverse Engineering of C / C++ Reverse Engineering of Perl Kieker in Space 5 Summary and Future Work W. Hasselbring (CAU) Kieker Southampton 0 / 5

11 Framework Features & Extension Points Essential Characteristics [Rohr et al. 008, van Hoorn et al. 009b; 0] Kieker Monitoring & Analysis Framework Modular, flexible, and extensible architecture (Probes, records, readers, writers, filters etc.) Pipes-and-filters framework for analysis configuration Distributed tracing (logging, reconstruction, visualization) Low overhead (designed for continuous operation) Evaluated in lab and industrial case studies Kieker is open-source software (Apache License, V..0) Recommended Tool of the SPEC Research Group Kieker is distributed as part of SPEC RG s repository of peer-reviewed tools for quantitative system evaluation and analysis, Monitoring Log/Stream Kieker.Analysis Kieker.Monitoring Sampling JMX Interface Monitoring Probe Adaptive Monitoring AnalysisTConfiguration Analysis / Visualization Plugin Analysis Controller Logging MonitoringT Controller MonitoringT Writer Monitoring Record Monitoring Reader Time/Source MonitoringT Probes/Samplers TimeT Sources MonitoringTT Writers MonitoringT Records MonitoringTT Readers Analysis/VisualizationT Plug-Ins Control/flow tracing Resource monitoring NamedDpipe <custom/monitoring/writer> OperationDexecution Control/flowDevents CPUDutilization ResourceDutilization CurrentDtime <custom/monitoring/record/type> FileDsystem JavaDMessagingDServiceDMJMS( JavaDManagementDExtNDMJMX( DatabaseDMSQL( NamedDpipe <custom/monitoring/reader> PipesDandDfiltersDframework ArchitectureDreconstrN Manual Servlet <custom/technology> TraceDanalysis Sigar Visualization CPUDutilization MemoryDusage <custom/monitoring/probe> SystemDtime FileDsystem CXFbSOAP AspectJ DatabaseDMSQL( JavaDMessagingDServiceDMJMS( JavaDManagementDExtNDMJMX( MemorybswapDusage TCPbIP PerformanceDAnalysis OutputDstream Real/timeDreplayer DependencyDgraphs SequenceDdiagrams CallDgraphs Servlet VB6XDCOBOLXDNNN <custom/interception//technology> <custom/time/source> TCPbIP <custom/visualization> <custom/trace/analysis> Spring <custom/reconstruction/plug-in> <custom/analysis/plug-in/tool> W. Hasselbring (CAU) Kieker Southampton / 5

12 Overhead Evaluation Linear Rise of Overhead [Waller and Hasselbring 0] Kieker Monitoring & Analysis Framework Execution time (median in µs) Writing (W) Collecting (C) Instrumentation (I) Method Time (T) 7, Slope ( ) ,8 5,7 4, 30,0,7 3, 5,0 0.,0 7,6 6,8 0, 0,7,5 3, Recursion depth (number of successive method calls) Experiment similar to: A5 AsyncFS writer 6 cores whole system is available MSEPT 0 Mul core Influence on Applica on Level Monitoring Overhead Jan Waller and Wilhelm Hasselbring W. Hasselbring (CAU) Kieker Southampton / 5

13 Kieker: Dynamic Analysis Workflow [van Hoorn et al. 0] Kieker Monitoring & Analysis Framework Analysis Configuration (via API and WebGUI) Monitoring Records Analysis Pipes and Filters Measurement Monitoring log/stream Invocations/minute [x 000] Workload Anomaly Detection searchbook() 0.5 getbook(..) 0.0 getoffers() <<execution container>> SRV0 $ 635 searchbook() 635 getbook(..) Calendar time @3:..Bookstore Monitoring Probe 573 getbook(..) 06 <<execution container>> SRV getoffers() getbook(..) Software System with Monitoring Instrumentation W. Hasselbring (CAU) Online and Offline Visualization Kieker Southampton 3 / 5

14 Core Kieker Framework Components Kieker Monitoring & Analysis Framework Kieker.Monitoring Monitoring Probe Monitoring Record Java probes/samplers: Monitoring Probes/Samplers Control-flow tracing Resource monitoring Manual instrumentation AspectJ Servlet Spring CXF/SOAP <your interception technology> Servlet Sigar CPU utilization Memory usage <your technology> <your monitoring probe> + basic adapters for C#/.NET Visual Basic 6/COM COBOL Monitoring Records Operation execution Control-flow events CPU utilization Resource utilization Current time Memory/swap usage <your monitoring record type> W. Hasselbring (CAU) Kieker Southampton 4 / 5

15 Core Kieker Framework Components Kieker Monitoring & Analysis Framework Kieker.Monitoring Time Source Monitoring Log/Stream Kieker.Analysis Monitoring Probe Logging Monitoring Controller Monitoring Writer Monitoring Record Monitoring Reader Periodic Sampling JMX Interface Monitoring Readers/Writers File system Java Messaging Service (JMS) Java Management Ext. (JMX) Database (SQL) Named pipe <your monitoring reader/writer> W. Hasselbring (CAU) Kieker Southampton 4 / 5

16 Core Kieker Framework Components Kieker Monitoring & Analysis Framework Kieker.Monitoring Time Source Monitoring Log/Stream Kieker.Analysis Pipe & Filter Configuration Monitoring Probe Logging Monitoring Controller Monitoring Writer Monitoring Record Monitoring Reader Analysis / Visualization Plugin Analysis Controller Periodic Sampling JMX Interface Analysis/Visualization Plugins Pipe-and-filter framework Architecture reconstr. Trace analysis Visualization Dependency graphs Sequence diagrams Call graphs <your visualization> <your trace analysis> <your reconstruction plugin> <your analysis plugin/tool> W. Hasselbring (CAU) Kieker Southampton 4 / 5

17 Program Instrumentation (Here: Manual) Example: Monitoring Operation Executions Kieker Monitoring & Analysis Framework Application code: Monitoring probe code (schematic): public void getoffers() { // EXECUTION to be monitored: catalog.getbook(false); } // BEFORE execution to be monitored if (!ismonitoringenabled()) { collectdatabefore(); } // AFTER execution to be monitored if (!ismonitoringenabled()) { collectdataafter(); writemonitoringdata(); } Instrumentation Getting the monitoring probe into the code Manual instrumentation Aspect-oriented programming (AOP), middleware interception,... W. Hasselbring (CAU) Kieker Southampton 5 / 5

18 description of the respective aspect probemethod, Tpmon saves AOP-Based Instrumentation by the Annotation current time before and after b is executed. Kieker Monitoring & Analysis Figure Framework.7: Sample system instrumented with Tpmon (a) and how an annotated operation is woven (b). aspect weaver basic functionality program aspect programming languages woven output code aspect description programs Figure.8: An aspect weaver weaves the aspects and the functional part of an application into a single binary (following (Kiczales et al., 3 public void getoffers() { 4 4 catalog.getbook(false); 5 } 6} Annotation-based (AOP) instrumentation for monitoring trace information W. Hasselbring (CAU) Kieker Southampton 6 / 5

19 Full Instrumentation with Kieker Kieker Monitoring & Analysis Framework Listing : META-INF/aop.xml <!DOCTYPE aspectj PUBLIC "-//AspectJ//DTD//EN" "http://www.aspectj.org /dtd/aspectj 5_0.dtd"> <aspectj> <weaver options=""> <include within="*"/> </weaver> <aspects> <aspect name="kieker.monitoring.probe.aspectj.operationexecution. OperationExecutionAspectFull"/> </aspects> </aspectj> Start the monitored application: $ java -javaagent:lib/kieker-.8_aspectj.jar BookstoreStarter W. Hasselbring (CAU) Kieker Southampton 7 / 5

20 Analysis Configuration with Kieker Example Pipe-and-Filter Configuration Kieker Monitoring & Analysis Framework operationexecutions <<Filter>> : Performance anomaly filter anomalyratings <<Reader>> : FS reader outputport systemmodel anomalyratings <<Filter>> : Anomaly graph plotter <<Repository>> : System model repository Workload Anomaly Detection Invocations/minute [x 000] Anomaly score Calendar time (hh:mm) W. Hasselbring (CAU) Kieker Southampton 8 / 5

21 Textual Analysis Configuration Kieker Monitoring & Analysis = = "eventcount", eventtypes = { Long.class })}) public final class CountingFilter extends AbstractFilterPlugin = "inputevents", eventtypes = { Object.class }) public final void inputevent(final Object event) { final Long count = this.counter.incrementandget(); super.deliver("eventcount", count); } } W. Hasselbring (CAU) Kieker Southampton 9 / 5

22 Graphical Analysis Configuration Web User Interface [Ehmke 03] Kieker Monitoring & Analysis Framework W. Hasselbring (CAU) Kieker Southampton 0 / 5

23 Web Control Center Cockpit [Ehmke 03] Kieker Monitoring & Analysis Framework W. Hasselbring (CAU) Kieker Southampton / 5

24 What others are doing with Kieker Analysis of Calling Networks [Zheng et al. 0] Kieker Monitoring & Analysis Framework Community structure Layered structure (hierarchy) Xi an Jiaotong University, Shaanxi [Zheng et al. 0] W. Hasselbring (CAU) Kieker Southampton / 5

25 akes it possible to access the facts using data visualisation and data analysis chniques, Visualizing hence obtaining Architecture a higher level Warehouses of understanding of those facts, that is taining knowdledge. In this section I deepen the definition by several examples. Example: Analyzing the JUnit Project [Dabrowski 0] Kieker Monitoring & Analysis Framework Fig.. Graph-based visualisation of artifacts in junit project Warsaw University [Dabrowski 0] W. Hasselbring (CAU) Kieker Southampton 3 / 5

26 Agenda Kieker Monitoring & Analysis Framework Dynamic Software Analysis Kieker Monitoring & Analysis Framework 3 Application Performance Monitoring 4 Reverse Engineering 5 Summary and Future Work W. Hasselbring (CAU) Kieker Southampton 4 / 5

27 Observations in field Extensive infrastructure monitoring, application monitoring not widespread Reactive monitoring probe injection only (after a critical performance drop has occurred) Monitoring on different system layers Application Performance Monitoring Application monitoring Infrastructure monitoring Business/service monitoring Business Processes Services Application Middleware Container Virtual Machine Operating System Hardware Key performance indicators, e.g. process throughput, SLO appliance, workload, Response times, operational profile, Thread/connection pool sizes, Heap size, CPU/memory utilization, Availability, reliability, Monitoring practice in the real world (based on what we ve seen) Focus on system level (network availability, resource utilization) or business level (key performance indicators) No systematic instrumentation on application level Monitoring as an afterthought : probes are only added when problems occurred. 3 / 5 W. Hasselbring (CAU) Kieker Southampton 5 / 5

28 Application-Level Monitoring in Practice...!? Among Java Professionals Application Performance Monitoring dynatrace Diagnostics dynatrace Quest Foglight CA/Wily Introscope IBM IT CAM Symantec Indepth for JEE None 4.0% 0.9% 5.8% 3.% 0.4% 65.5% Others n.s. 4.3% 9.9% 0% 0% 0% 30% 40% 50% 60% 70% 80% Java monitoring tools being used Java monitoring largely unknown. [codecentric GmbH 009] W. Hasselbring (CAU) Kieker Southampton 6 / 5

29 Application-Level Monitoring in Practice...!? At Facebook Application Performance Monitoring Scaling Facebook to 500 Million Users and Beyond Making lots of small changes and watching what happens only works if you re actually able to watch what happens. At Facebook we collect an enormous amount of data any particular server exports tens or hundreds of metrics that can be graphed. This isn t just system level things like CPU and memory, it s also application level statistics to understand why things are happening. It s important that the statistics are from the real production machines that are having the problems, when they re having the problems the really interesting things only show up in production. The stats also have to come from all machines, because a lot of important effects are hidden by averages and only show up in distributions, in particular 95th or 99th percentile. Robert Johnson, Facebook Engineering Director W. Hasselbring (CAU) Kieker Southampton 7 / 5

30 Anomaly Detection + Diagnosis Xing AG [Bielefeld 0, Frotscher 03] Application Performance Monitoring W. Hasselbring (CAU) Kieker Southampton 8 / 5

31 Anomaly Detection + Diagnosis Previous work [Marwede et al. 009] Application Performance Monitoring presentation.orderbean [ 454/397 0,094 7,8% ] service.hessian.client.orderservice [ 484/98 0,057 6,9% ] Virtual Machine klotz $ org.apache.struts.action.actionservlet [ 487/ ,90 7,8% ] presentation.cartbean [ 07/70 0,087 5,99% ] service.hessian.client.catalogservice getproductlistbycategory(string) [ 6365/437 0,6 0,006 3,4% ] presentation.catalogbean [ 838/6837 0,048 6,88% ] presentation.accountbean [ 494/088 0,06 6,6% ] 088 service.hessian.client.accountservice [ 53/088 0,033 6,35% ] getproduct(string) [ 369/967 0,046 0,8,7% ] getproductlistbycategory(string,int,int) [ 640/437 0,6 0,06,8% ] 437 getitemlistbyproduct(string) [ 967/967 0,995 0,406 4,43% ] 967 getitemlistbyproduct(string,int,int) [ 967/967 0,995 0,488 4,69% ] getcategory(string) [ 6309/349 0,6 0,037 3,04% ] service.hessian.client.orderservice [ 484/98 0,057 6,9% ] service.hessian.client.accountservice [ 53/088 0,033 6,35% ] Virtual Machine tier [ 447/6098 0,03 5,90% ] Virtual Machine scooter [ 88/76 0,07 3,43% ] Virtual Machine puck [ 447/943 0,03 4,48% ] Component Level Operation Level Deployment Context Level Component Level service.hessian.client.catalogservice getproduct(string) [ 369/967 0,046 0,8,7% ] getproductlistbycategory(string) [ 6365/437 0,6 0,006 3,4% ] getitemlistbyproduct(string) [ 967/967 0,995 0,406 4,43% ] getcategory(string) [ 6309/349 0,6 0,037 3,04% ] Component Level Operation Level getproductlistbycategory(string,int,int) [ 640/437 0,6 0,06,8% ] getitemlistbyproduct(string,int,int) [ 967/967 0,995 0,488 4,69% ] 437 W. Hasselbring (CAU) Kieker Southampton 8 / 5

32 Anomaly Detection + Diagnosis ExplorViz [Fittkau et al. 03] Application Performance Monitoring W. Hasselbring (CAU) Kieker Southampton 8 / 5

33 Anomaly Detection + Diagnosis Work at Stuttgart University on online failure prediction [Pitakrat et al. 04] Application Performance Monitoring Component-level Predictors PCM SLAstic... HDD Failure Predictor CDT PAD System-level Predictor Monitoring Reader!! Event Log Analyzer Hora Kieker, Weka, R, PRISM, ESPER,... W. Hasselbring (CAU) Kieker Southampton 8 / 5

34 Agenda Reverse Engineering Dynamic Software Analysis Kieker Monitoring & Analysis Framework 3 Application Performance Monitoring 4 Reverse Engineering Reverse Engineering of Java EE Reverse Engineering of C# Reverse Engineering of Visual Basic 6 Reverse Engineering of COBOL Reverse Engineering of C / C++ Reverse Engineering of Perl Kieker in Space 5 Summary and Future Work W. Hasselbring (CAU) Kieker Southampton 9 / 5

35 Extending the Pipe-and-Filter Configuration Here: for Reverse Engineering Reverse Engineering getbook(..) <<Reader>> : FS reader traceevents outputport <<Filter>> : Trace reconstruction filter <<Repository>> : System model repository systemmodel messagetraces executiontraces messagetraces messagetraces systemmodel <<Filter>> : Sequence diagram visualization systemmodel <<Filter>> : Dependency graph visualization searchbook() getbook(..) getoffers() <<execution container>> SRV0 getbook(..) 635 searchbook() getoffers() Kieker.Analysis example pipes-and-filters configuration Performance anomaly detection and visualization Architecture and trace reconstruction/visualization 06 <<execution container>> getbook(..) W. Hasselbring (CAU) Kieker Southampton 30 / 5

36 Generated Sequence Diagram Reverse Engineering W. Hasselbring (CAU) Kieker Southampton 3 / 5

37 Reverse Engineering of Java EE Customer portal at EWE TEL GmbH Reverse Engineering Reverse Engineering of Java EE Distributed Enterprise Java System instrumented subsystem <<device>> Front End Server (on site) <<device>> Client <<device>> 0.. H/W LB * * <<execution environment>> Web Server/ Java EE Container <<artifact>> PortalServiceA <<artifact>> PortalServiceB <<artifact>> PortalServiceC <<device>> Application Server <<execution environment>> Java EE Container <<artifact>> BusinessServices M N <<device>> DB Cluster <<device>> Back End System <<device>> 3rd Party System 0.. <<device>> Front End Server (off site) K Servlet, Spring and CXF/SOAP probes W. Hasselbring (CAU) Kieker Southampton 3 / 5

38 Software Architecture Level Component dependency graph Reverse Engineering Reverse Engineering of Java EE $ <<execution container>> css0 <<execution container>> css <<execution container>> pikdb0 <<execution container>> pikdb W. Hasselbring (CAU) Kieker Southampton 33 / 5

39 System Architecture Level Component dependency graph Reverse Engineering Reverse Engineering of Java EE $ <<execution dofilter(..) <<execution dofilter(..) handlemessage(..) handlemessage(..) handlemessage(..) <<execution container>> pikdb <<execution container>> pikdb isuptodate(..) getzone(..) deleteresourcerecord(..) addresourcerecord(..) getwebuserids(..) getwebhostingcategories(..) updateftpaccount(..) updatemysql(..) terminateftpaccount(..) createftpaccount(..) changedomainusagecategory(..) createwebspace(..) getwebspacesubdirectories(..) getwebhostingcategory(..) getprowebdomainnames(..) getwebhostingdetails(..) getwebspacequota(..) get frommasteraccount(..) createwebstatssession(..) getavailablephpversions(..) addsmscontact(..) sendsupportrequest(..) getsmsauthorizations(..) getftpquota(..) getincreasebandwidthresult(..) getorderstaticipresult(..) getmailquota(..) getsmscontacts(..) sendsms(..) getantivirlicenselist(..) getsmsnumbers(..) removesmscontact(..) queuesms(..) getvirtualserverenvironment() setsmsauthorizations(..) getsmsauthorization(..) createwlanaccount(..) addantivirlicense(..) getmysqlversion(..) activatesmsnumber(..) createsmstoken(..) scheduleantivirlicenseactivation(..) isdomainfree(..) getdomains(..) gettopleveldomains(..) getdomains(..) reassigndomain(..) createsubdomain(..) restartdomaintransfer(..) getfederalstateforzipcode(..) getdomaininfo(..) getdomaincategory(..) performdomainorder(..) updatevacationinfo(..) getmailaccounts(..) update forward(..) getvacationinfo(..) delete virusscan(..) add virusscan(..) updatespamfilterconfiguration(..) removelistentryfromspamfilter(..) create account(..) add alias(..) delete alias(..) getspamfilterconfiguration(..) addlistentrytospamfilter(..) getcustomeraccount(..) checklogin(..) acknowledgednstermsofuse(..) getmyadminloginlink(..) hasacknowledgeddnstermsofuse(..) getaccountnumberfordbuser(..) passwordrequestbymailallowed(..) updateaccountpassword(..) deletetoken(..) updatecsspassword(..) createtoken(..) checktoken(..) getenddatefrommaincontract(..) getpackageconfiguration(..) modifyresourcerecord(..) isuptodate(..) deleteresourcerecord(..) addresourcerecord(..) getwebspacesubdirectories(..) createwebspace(..) getwebhostingcategory(..) getavailablephpversions(..) get frommasteraccount(..) createwebstatssession(..) getwebspacequota(..) getwebhostingdetails(..) getwebuserids(..) getwebhostingcategories(..) hasproweb(..) createmysql(..) updatewebspace(..) setphpversion(..) createftpaccount(..) createwebspacesubdirectory(..) updateftpaccount(..) getprowebdomainnames(..) scheduleantivirlicenseactivation(..) sendsupportrequest(..) addantivirlicense(..) getincreasebandwidthresult(..) getorderstaticipresult(..) queuesms(..) getmailquota(..) getsmsauthorizations(..) getantivirlicenselist(..) getftpquota(..) getsmsauthorization(..) sendsms(..) getsmsnumbers(..) getsmscontacts(..) activatesmsnumber(..) createsmstoken(..) getvirtualserverenvironment() createwlanaccount(..) orderstaticip(..) setsmsauthorizations(..) 45 3 getmysqlquota(..) 3 removesmscontact(..) addsmscontact(..) getdomains(..) gettopleveldomains(..) createsubdomain(..) reassigndomain(..) performdomainorder(..) isvalidcontact(..) deletesubdomain(..) isdomainfree(..) getdomaininfo(..) getfederalstateforzipcode(..) getdomaincategory(..) getvacationinfo(..) getmailaccounts(..) updatevacationinfo(..) reactivate virusscan(..) delete virusscan(..) add virusscan(..) removelistentryfromspamfilter(..) add alias(..) delete account(..) addlistentrytospamfilter(..) updatespamfilterconfiguration(..) create account(..) getspamfilterconfiguration(..) update forward(..) createtoken(..) getenddatefrommaincontract(..) getpackageconfiguration(..) getcustomeraccount(..) getaccountinfos(..) checklogin(..) getwebstatisticlink(..) getaccountnumberfordbuser(..) acknowledgednstermsofuse(..) hasacknowledgeddnstermsofuse(..) getmyadminloginlink(..) printcontractconfirmation(..) updatecsspassword(..) updateaccountpassword(..) checktoken(..) deletetoken(..) <<execution container>> <<execution container>> css $ 5300 <<execution container>> css <<execution container>> @5:..AccountService W. Hasselbring (CAU) Kieker Southampton 34 / 5

40 Dynamic Analysis for Modernization KoSSE Project DynaMod, Reverse Engineering. Reverse Engineering of C# Motivation: Emergent Architectures Project Consortium Dataport, Altenholz Software Engineering Group University of Kiel, Kiel b+m Informatik AG, Melsdorf HSH Nordbank AG, Kiel W. Hasselbring (CAU) Kieker Southampton 35 / 5

41 Reverse Engineering of C# Complete Test Suite of Nordic Analytics [Magendanz 0] Reverse Engineering Reverse Engineering of C# Case study at HSH Nordbank AG. W. Hasselbring (CAU) Kieker Southampton 36 / 5

42 After Selecting a single Use Case Reverse Engineering of C# Reverse Engineering Reverse Engineering of C# $ @5:..cSecurityDetailsCFP @48:..cConvexityAdjustmentModel @3:..cSecurityDependency @3:..cVaRMethodFactory @4:..cPricingParamsCFP @50:..cAccruedInterestCalculator W. Hasselbring (CAU) Kieker Southampton 37 / 5

43 Zoom to the Operation Level Reverse Engineering of C# Reverse Engineering Reverse Engineering of IsShortRateMC(..) CreateSwaptionPricing(..) Price(..) InitializeLMMPricing(..) InitializeCommodityMCPricing(..) InitializeIndexMCPricing(..) IsHistoricalSecurity(..) AddIPricingResults(..) GetPricingResult(..) Price(..) $ 4 PriceSwaption(..) SetContactMessage() ReadObjectsFromFile(..) WriteObjectsToFile(..) ClearPools() EvaluateJob(..) GetVersionNumber() GetCompilationDate() 6 SetResultProperties(..) GetObject(..) SetJobPart(..) PriceSwaption(..) GetVolatility(..) AddResultProperty(..) AddResultProperty(..) CreateObject(..) CreateObject(..) GetOutConverter(..) WriteLogInfo(..) WriteLogInfo(..) InitializeTemplateObjects() SetStoreLossDistributions() InitializeUsageTracking() InitializePartialJobs(..) GetUseUsageTracking(..) Evaluate() InitializeMarketDataUsageTracking() GetValuationDate() <<assembly 8 InitializeObjectTracking(..) GetObject(..) MarketDataUsageCheck() BuildNotBuiltDefaultCurves() ClearIntermediateObjects() ToString() BuildDateGrids(..) GetSwaptionValueEvaluationType(..) GetBlackScholesOptionType(..) StoreAdditionalPricingResults(..) FairRate(..) CreateObject(..) CreateObjectPreProcessing(..) GetObjects(..) Objects(..) JobPartObjects(..) ReplaceDependence(..) Vega(..) PriceFixLeg(..) InitializeMethodAssetClassSpecification() ReadMethodAssetClassSpecification(..) GetDefaultIRSubModelType(..) Create(..) ReadObjectsTable(..) InitMembersWithCorrespondingNullValues() Initialize(..) 4 InitializeObjects() CopyOrId(..) FindCollections(..) SetCollectionObjects(..) GetObjectSignatures(..) CheckObjectsAreUnique(..) GetObjects(..) Objects(..) GetEndDate() GetSwapTenor() PriceFixCF(..) ReadDescriptionTableProperties(..) ReadObjectsTable(..) CheckNumberOfObjects(..) ProveOfUniqueObjects(..) ReadObjects(..) Insert(..) 4 8 ReadDescriptionTableProperties(..) GetSwapTenorTime() GetHashCode() ReadCurrencyProperty(..) GetDependencyHandle() ReplaceDependence() GetObject(..) PriceSwaption(..) GetSpecificMarketData(..) PriceSwaption(..) GetVolatility(..) GetPricingResult(..) Price(..) SetResultProperties(..) AddResultProperty(..) AddResultProperty(..) AddIPricingResults(..) Price(..) 4 W. Hasselbring (CAU) Kieker Southampton 38 / 5

44 Reverse Engineering of Visual Basic 6 Analysis of Specific Traces Reverse Engineering Reverse Engineering of Visual Basic 6 frmlogin.cmdok_click() frmlogin.anwendung() frmmain.mdiform_load() Module.Main() frmlogin.form_load() frmakte.tool_menu_begin() frmlogin.txtpassword_gotfocus() frmakte.ob_frei() $ frmmain.mnudatafrmakte_click() frmakte.form_resize() frmakte.form_load() frmakte.ob_sperren() frmakte.felder_sperren() frmakte.option_click(integer) clsklasse.class_initialize() clsklasse.pruef_user() clsklasse.pruef_abt() clsklasse.pruef_rechte() BENUTZER.select() ABTEILUNG.select() BENABT.select() frmakte.felder_frei() frmakte.setbuttons(boolean) AKTE.select() clsklasse.adoprimaryrs_movecomplete(adodb.eventreasonenum, ADODB.Error, ADODB.EventStatusEnum, ADODB.Recordset) frmakte.suchen() frmakte.felder_sichern() 8 clsklasse.class_getdatamember(string, Object) frmakte.cmdedit_click() frmakte.indexposlesen() IPOS.select() frmakte.pruefung_ob() frmakte.fuehrende_nullen(long) frmakte.txtfields_change(integer) frmakte.daten_binden() 6 frmakte.tool_menu_edit() frmakte.abteilung() frmakte.rtxtboxtitel_change() frmakte.rtxtboxenthaelt_change() Case study at Dataport W. Hasselbring (CAU) Kieker Southampton 39 / 5

45 cmdupdate_click mnufileclose_click Identification of unused Functions Reverse Engineering of Visual Form_KeyDown Basic 6 Reverse Engineering Reverse Engineering of Visual Basic 6 mnueditundo_click SetButtons cmdnext_click frmstat.847 ms cmdlast_click cmdfirst_click cmdclose_click 85 Abt_Akte_del Ben_BenAb ipos_loesch Statistik 6.7 ms ms Abt_Zus_del 76 Ausgabe cmd_start_click abt_suchen ms Form_Load MouseWheel mnufilestart_click ms txt_abt_gotfocus Zeile mnuhelpabout_click MSHFlexGrid_RowColChange EditingRecord Itext_neu Add IndexNeu mnuhelpsearchforhelpon_click 3.9 ms Zus_Akte_del ms Form_KeyDown Ben_neu Abt_n cmdclose_click 83 mnuhelpcontents_click mnufileexit_click cmd_druck_click AkteLoeschen Zus_Itext_del W. Hasselbring (CAU) Kieker Southampton 40 / 5

46 Reverse Engineering of COBOL With consideration of non-instrumentable modules [Knoche et al. 0] Reverse Engineering Reverse Engineering of COBOL IDENTIFICATION DIVISION. PROGRAM-ID. MODULE-X. PROCEDURE DIVISION. CALL "MODULE-Y". GOBACK. Literal block Before-Call Injection Point Called Module: "MODULE-Y" Literal block AOP-based COBOL instrumentation MODULE-B $ MODULE-A MODULE-C MODULE-D MODULE-E Module-level call dependency graph with assumed dependencies W. Hasselbring (CAU) Kieker Southampton 4 / 5

47 Analysis of a PC COBOL System [Richter 0] Reverse Engineering Reverse Engineering of COBOL Abbildung 6.5: Komponentenabhängigkeits-Modell der Menü-Navigation; Extrahiert aus den Monitoringdaten @8:REALIA_FINDCLOSE Ergebnisse der Evaluation W. Hasselbring (CAU) Kieker Southampton 4 / 5

48 C++ Digital Signal Processing Kiel Real-time Audio Toolkit (KiRAT) Reverse Engineering Reverse Engineering of C / C++ Source: W. Hasselbring (CAU) Kieker Southampton 43 / 5

49 on_main, algcore_common_audio_effects_chorus_main, algcore_common_wavreader_main, algcore_common_fb_synthesis_main, algcore_common_wavreader_interface, algcore_common_si_main _conv_ols04] _matrix_cpu_main. Analysis of und the algorithmic kernel [Mahmens gnalsel algcore_common Reverse Engineering of C++ Reverse Engineering. Reverse Engineering of C / SapAlgCoreFree3DArray(..) SapAlgCoreFree(..) SapAlgCoreCalloc3DArray(..) SapAlgCoreMedsigproDeinit(..) SapAlgCoreCalloc(..) 3 *SapAlgCoreMedsigproInit(..) 3354 SapAlgCoreFreeDArray(..) SapAlgCoreCallocDArray(..) SapAlgCoreFreeDArray(..) SapAlgCoreCallocDArray(..) Performance normierte Laufzeit der Funktion [in µs] Abbildung 5.4. Aufrufe des Pakets Medical-Signal-Processing zur Speicherverwaltung Abbildung 5.4 zeigt die Nutzung der Speicheralloziirung des Pakets. Wie auch beim Paket System-Evaluation wird die Funktion SapAlgCoreCallocDArray() einmal weniger aufgerufen als die Funktion SapAlgCoreFreeDArray(), Aufrufe im Vergleich zu 3 Aufrufen. Alle anderen Ressourcen werden vollständig wieder freigegeben normierte Zeit [in s] W. Hasselbring (CAU) Kieker Loudspeaker-Enclosure-Microfone-Simulation Abbildung 6.4. Aggregiertes Sequenzdiagramm für die Laufzeiten der Funktion SapAlgCorePro- Southampton 44 / 5

50 Reverse Engineering of Perl Visualizing Dependencies in EPrints Reverse Engineering Reverse Engineering of Perl First shot with full instrumentation: W. Hasselbring (CAU) Kieker Southampton 45 / 5

51 Visualizing Dependencies 6. Ergebnisse in EPrints Customized visualization with Gephi [Wechselberg 03] Reverse Engineering Reverse Engineering of Perl Abbildung 6.. Visualisierung der Abhängigkeiten in EPrints Red: EPrints.PluginFactory. Green: EPrints.Repository. W. Hasselbring (CAU) Kieker Southampton 46 / 5

52 Visualizing Dependencies in EPrints Refinement with adaptive monitoring [Jensen 03] Reverse Engineering Reverse Engineering of Perl Abbildung 5.6. Abhängigkeiten beim ersten Request mit dem adaptiven Monitoring von EPrints. W. Hasselbring (CAU) Kieker Southampton 47 / 5

53 Kieker in Space Some analysis results [Harms 03] Reverse Engineering Kieker in Space W. Hasselbring (CAU) Kieker Southampton 48 / 5