User Guide - CloudScale Project
Contents
1. The Extractor is a reverse engineering tool for automatic model extraction It parses source code and generates partial ScaleDL models that are further used by the Analyser and the Static Spotter nn nn nn nn nn nn nn nnn ne enn een nnn Analyser r Spotter Static spotter Dynamic spotter Results Results B l Problem There are situations for a software system that the original architecture has been lost or outdated due to further development The software architect has no detailed picture of what the software system s architecture looks like but only the source code While the software architect would require the architecture when migrating the software system to a new environment such as cloud computing environment ll Features The Extractor is based on the Archimetrix approach Archimetrix is a tool supported reengineering process that combines different reverse engineering approaches to enable an iterative recovery and reengineering of component based software architectures The Extractor is focused on parsing and architecture reconstruction The Extractor parses the source code extracts and clusters a component based software architecture based on source code metrics The software architecture is generated as a partial ScaleDL model l Input amp Output As for an input the Extractor takes Java source code and generates partial ScaleDL model as an output2. To create an Overview model instance user has to create a new Overview alternative This can be achieved by double clicking on the Overview section in the Workflow diagram or by using the Dashboard editor In the Dashboard editor user interface section for manipulating Overview alternatives is under the Overview tab item Create button opens up the wizard for creating an empty Overview alternative or creating initial model from the Extractor result or external partial PCM model Extended Palladio Component Model Extended PCM Extended PCM allows architects to model the internals of the services components components assembly to a system hardware resources and components allocation to these resources the extension allows additionally to model self adaptation monitoring specifications and adaptation rules Relation to Workflow An instance of the Extended Palladio Component Model Extended PCM is part of a ScaleDL instance the input of the Analyzer see Workflow Features A PCM instance allows software architects to model the internals of services The PCM allows to model components components assembly to a system hardware resources components allocation to these resources and static usage scenarios The extended refers to additional models added in the CloudScale context These models cover monitoring specifications service level objectives and self adaptation rules Monitoring specifi
3. Architectural Templates ATs ScaleDL Architectural Template allows architects to model systems based on best practices as well as to reuse scalability models specified by architectural template engineers Il Relation to Workflow Architectural Templates ATs are part of ScaleDL the input of the Analyzer see Workflow Also the transformation from Overview model to Analyzer input utilizes ATs to simplify the transformation specification ll Features Architectural Templates ATs help software architects to specify Extended PCM instances more efficiently based on reusable ScaleDL templates We provide the CloudScale AT catalogue that includes best practice templates for designing and analyzing scalable elastic and cost efficient SaaS applications We based this catalogue on common architectural styles and architectural patterns found in cloud computing environments Input amp Output For an AT based design software architects need a catalogue of available ATs as an input For cloud computing applications we suggest the CloudScale AT catalogue The output of an AT based design is part of a normal ScaleDL instance The Analyser fully supports AT enabled ScaleDL instances Walkthrough Software architects conduct the following actions for designing ScaleDL instances with ATs 1 Switch either to the CloudScale Perspective or the Analyser Perspective using the menu bar buttons
4. a Goto Eclipse gt Help gt Install New Software b Add CloudScale Toolchain update site http cloudscale xlab si cse updatesites toolchain nightl c Install Toolchain features Analyser Extractor Static Spotter and Dynamic Spotter Dependencies and Sources sometimes Dependencies needs to be installed first and after that everything else 3 Clone repository a Sgit clone https github com CloudScale Project Environment git 4 Import CloudScale Environment plugins under plugins directory into the workbench 5 Run product eu cloudscale env product Introduction The CloudScale Environment CSE is an open source solution oriented to provide an engineering approach for building scalable cloud applications by enabling analysis of scalability of basic and composed services in the cloud It is a desktop application integrating CloudScale tool chain consisting of Dynamic and Static Spotters the Analyzer and the Extractor while driving the user through the flow of the CloudScale Model Application can be installed and used in any personal computer running Java 6 including Windows MacOS and Linux Environment perspective Environment perspective is the main perspective in the CSE responsible to provide all main functionality of the CloudScale toolchain through unified and well defined components It can be accessed through Tools gt Environment gt Perspective action or through toolbar action first button In this pe
5. Walkthrough 1 Click the Extractor on Dashboard go to Run 2 Click Create Create a new configuration Project Worktow gegen sx Aval Rate Piot 1t CoudScsie Fie Eat Tools Story Diagram Hsp BR OSA BSE Bop mm mamm A erg Sousscale D Project Eege El ES 7 D Dashbord ProdueerConmumer 77 Oashboard Ovo El o ee aia teatro rami Seon sac otr e Currey are are no ernest clay Crosse new cow Crome new Project eut Bagram sx Acrival Rate Piot rt FPO CSL AA 3 Select input project configure the metrics 4 Run Clustering Composition Threshold Min End Value 25 Clustering Composition Threshold Max Start Value 100 Clustering Composition Threshold Decrement dzee Clustering Merge Threshold Min Start Value Clustering Merge Threshold Max End Value 5 Click Results tab under Main menu you can see a bunch of models have been generated Model JavaMpplcaon Resource ProducerConsumer ava gt Model Repository aName Resource internal_architecture_model repository Model System SoMoX Reverse Engineered System Resource internal_architecture mod gt Model SourceCodeDecoratorRepository Resource internal_architecture_model sourcecod References Ref to CS Method Ref to Extractor User manual Introduction CloudScale s Analyser allows to a
6. D I Ss A Property Value Function Output At End 530 0 Function Output At Start 0 0 The LIMBO tool provides a tree based editor as shown in the figure At the root of the model is a Sequence which typically contains one or more TimeDependentFunctionContainer that can be regarded as segments of the sequence Each segment has a duration that adds to the total duration of the Sequence When using multiple LIMBO models to describe work parameters as well as load the total duration of each Sequence must be the same Each segment contains a single function e g linear exponential logarithmic or sin trend describing how the value evolves within that segment and typically contains a start and end value see e g properties for LinearTrend in the figure Note that the LIMBO tool originally was developed with focus on arrival time and thus uses arrival time in some of the labels e g the y axis in the graph However the same modeling concept can be use to express the evolution over time of other values such as population in closed workload or even work parameters such as average data size e g video or picture size See the LIMBO web pages for more documentation on the tool Creating a run configuration The next logical step is to create a usage evolution model but as this model is initialised along with the run configuration we describe creating a run configuration first To create a run configuration open to the
7. the Dashboard editor Overview model when completed can later be used to produce the input model for the Analyser see Extended Palladio Component Model l Features An Overview model instance allows software architects to model platform services software services connections between them and deployment in the cloud environment It includes hardware specifications for the different cloud providers Those specifications are designed in a way to support extending existing cloud specifications by the use of a system descriptors Overview meta model can be used as a general structure for describing software architecture inside cloud environment In CloudScale Environment its main purpose is to ease and accelerate the modeling of the Extended PCM model by the use of partial PCM to Overview and Overview to Extended PCM transformations Software services in Overview model can embed partial PCM models to describe internal mechanics When the Overview model is transformed to the Extended PCM partial PCM models are combined to form a complete input for the Analyser ll Input amp Output Overview model can be modelled with the Overview diagram editor Initial configuration can be made with the import wizard from the Extractor output model or external partial PCM model Repository and System The main output of the Overview model is the Extended PCM model which is used as the input for the Analyser see Analyser ll Minimal Example
8. Analyser tab of the Dashboard and the Run tab within it Click the create button in the Alternative conf column When no configuration has been added yet you can also use the Create new button that will be present in the main editing area on the middle right side of the screen For the usage evolution a Normal alternative type should be selected in the dialog that appears Under basic settings measurement count stop condition should be unchecked set to 1 The value of simulation time stop condition should to equal or higher than the Final Duration of the LIMBO models The details of setting up a run configuration is further described in the Analyser chapter Creating the Usage Evolution model The usage evolution model describes how a usage scenario of a usage model evolves over time and uses LIMBO models to define the evolution of the individual load and work parameters As described under Creating a run configuration a usage evolution model is created as part of a run configuration and will appear under the subfolder of your selected configuration alternative for the analyser The figure below shows the ScaleDL Usage Evolution tree based structure editor At the root of the model is a single usage evolution element To add content to the model select the root element and select New Child Usage from the context menu to create a Usage Then select the newly created Usage element its properties
9. Dynamic Spotter Tab Click Server under Main menu Start Built in Spotter server Click Connect Built in Spotter server External Spotter server Server port coo sewe Server is up and running gt Server Log Starting DynamicSpotter server port 8080 server to gt Users Jinying Downloads eu cloudscaleproject env master product itents MacOS workspace ds server macosx cocoa x86_64 1 2 Eclipse app Cont Builtin server persisted in Users Ji cloudscaleproject env master product macosx cocoa x86_64 1 2 Eclipse apy ds serverjar SERVER OUTPUT 07 14 16 147 main INFO org Ipe common util web WebServer Web Server started on port 8080 2 Click Input gt Create add gt Instrumentation and Measurement f QU respectively Su es d TT TE EEN 3 Run 4 Click Results the result is shown as follows
10. input configuration and results and these are analogous with the dashboard sections which are displayed by double clicking on the state In addition these states also contain a toolbar with available actions e g wizards quick run and a representation of required resources in an input state available run configurations in a run state and available results in a results state Workflow diagram has two types of connections required solid and optional dashed First defines that current state is active only when all required states are validated while the latter one does not expect the optional state to be validated however it normally represents tools interoperability and thus producing required resources quicker Entry points where user can start at the beginning are all the states that does not require another one to be fulfilled tools input states and all ScaleDL models Tools perspectives Complete CloudScale tool chain is available in the CloudScale Environment separately through dedicated tool s perspectives which can be accessed through Tools menu or directly using toolbar buttons From these perspectives tools can be used separately as a stand alone products without the need of using Environment integration functionalities For instructions how to use integrated tools separately please see Tools tutorials To open other views UI components that may be missing in the menus one can access them through Tool
11. ok or something needs to be done fixed Examples CloudScale Environment integrates many examples and templates including stand alone examples for each of the integrated tools Examples can be accessed through CloudScale Examples project in New Project wizard ScaleDL The Scalability Description Language ScaleDL is a language to characterize cloud based systems with a focus on scalability properties ScaleDL consists of five sub languages three new languages ScaleDL Usage Evolution ScaleDL Architectural Template and ScaleDL Overview and two reused language Palladio s PCM extended by SimuLizar s self adaption language and Descrartes Load Intensity Model DLIM In this section for each we describe how it is used in the CloudScald Evironment Overview The ScaleDL Overview is a meta model that provides a design oriented modelling language for cloud based system architectures and deployments It has been designed with the purpose of modelling cloud based systems from the perspectives of service deployments dependencies performance costs and usage It provides the possibility of modelling private public and hybrid cloud solutions as well as systems running on non elastic infrastructures Relation to Workflow An instance of the Overview model is a part of a ScaleDL instance It can be generated from the Extractor result partial PCM model or directly by creating a new Overview alternative in
12. problems in Cloud applications The tool consists of two main components that are provided as two separate programs Static and Dynamic spotters The Static Spotter analyses the application code to find scalability anti patterns The Dynamic Spotter analyse the execution of the application through systematic measurements to find scalability anti patterns in the application s behaviour Within this document we will refer to the pair as Spotter and to their individual components as either Dynamic or Static Spotter Walkthrough Below are steps and most basic information needed to start working with the CloudScale Environment 1 Start CloudScale Environment Open Environment perspective through menu Tools gt Environment gt Perspective 3 Create new CloudScale Project a b File gt New gt CloudScale Project Click on first toolbar item New CloudScale project 4 Go through available components a b Cc d Dashboard accessible through double clicking project cse file Workflow accessible through double clicking method workflow file ScaleDL models available under ScaleDL models folder Each Tool has has dedicated folder where all configurations and results are stored 5 Start creating input and run alternatives for the tool of your choice using Dashboard components Help with Workflow diagram which highlights steps that are available and shows validation information if everything is
13. will be shown in the Property editor tab at the bottom of the screen eoo CloudScale e Ss R Project Explorer 23 amp Y Dashboard ImageArchiveAnalyser Lo analyser usageevolution 3 O v 5 ImageArchiveAnalyser CloudScale Project Resource Set v Analyser v amp Input v Alternative_2015 07 03_02 25 04 Lei default allocation default allocation_diagram default repository gt S default repository_diagram gt LO default system f default system_diagram gt My resourceenvironment gt GB My resourceenvironment_diagram prop alt gt publishOpen usagemodel gt S publishOpen usagemodel_diagran gt Se Alternative_2015 07 03_02 45 35 gt lw Alternative_2015 07 03_02 50 03 gt X Configuration Results Selection Parent List Tree Table Tree with Columns v platform resource ImageArchiveAnalyser 20Analyser Configuration Alternati Y Usage Evolution aName Y Usage aName Work Parameter Evolution 5 Project workflow disgram BmageAr 38 El Properties 3 Validation Arrival Ra J Simulatio Console ei ra gt Y Property Value Selected Nothing To reference elements from the usage model and LIMBO models created in previous steps in the usage evolution they must first be imported into the usage evolution editor This is done by dragging the files into the editing area of the usage evolution editor or use Load resource from the context menu to locate th
14. 2 Open an existing PCM system model or created a new one see Extended Palladio Component Model Extended PCM 3 Open the system in a tree editor i e double click the system file not a system_diagram file 4 Right click on the system element of your system model and choose MDSD Profiles gt Apply Profile 5 Inthe new dialog select the AT you want to apply and apply it By default all ATs of the CloudScale AT catalogue are available If you want to use third party AT catalogues assure that you installed the respective plugins included ATs will then also be shown in this dialog 6 Assign all AT roles to respective elements of your ScaleDL instance For example if an AT specifies an AT role for PCM s assembly contexts right click this assembly context within the system model tree editor and select MDSD Profiles gt Apply Stereotype and apply the role similar to the previous action Our CloudScale AT catalogue describes per AT which roles have to be assigned 7 Aseach AT role can have a set parameters software architects also need to set such parameters when requested These parameters can simply be set via the Properties View for elements that have an AT role assigned Usage Evolution UE The purpose of the ScaleDL Usage Evolution is to enable modelling of how the load and work of a system evolves over time ScaleDL Usage Evolution provides a thin modelling layer that expresses how the
15. CloudScale Environment User Guide CF CloudScale Funded by the European Commission s Framework Programme Seven FP7 SINTEF GITT ERICSSON ZB xan hantega Contents A A aen aaea Ea aaa AEAEE aeaa Ea AE a aA A AE A aaa A E a Eia 5 rte ele Ee 5 OWN nieren ae E RE eE PEET RE Aai 5 Inst llation amp RUN eegene 5 Development it A AA EA a EEE AE A Ad it A 6 Maven Tycho auna acti 6 Eclipse IDE EE 6 Ha E 7 Environment Eet 7 CloudScale Environment PRO Odd 7 Dashboard A RC UE O E OA E E 8 A 10 Tools PETS PECES A RE EREEREER ERR ERENG 11 e EE 12 EXIME il 12 A encumber dtevan paca inieion eet ane en eeaaanies 13 OVEN LEE 13 EE to WV OV IOWA 13 FOOL ES la 13 INPUES QUEDU Eisner idad 14 Mi imal Example iia EA A A A RE NAR ERENER 14 Extended Palladio Component Model Extended DCH 15 REINO to WOrKfIOW Siria ee 15 ee 15 INPUT amp EIERE e dek ENER shine aida peas Sade dE ENER pes ces 15 Minimal Example A a ia 15 Architectural Templates ATS einen diia 17 Relation to WORMS A A A A ae a 17 AN NO 17 ele UE Te E 17 IVOIKEOT OIDO EE Error Bookmark not defined USAGE ele DEN ii E EE TA ROS SE EA ARES 19 Relation to WordlOWi eege Seege 19 AU ERA EA A EEE OEE i a 19 WIEN CUT Decade 19 EXTIOCLOL cdta ii airada s ean ies 24 ET O 24 Relation to Worktiow oia cationes 24 PEDIA IA OT IAL Otte ROPE Pen NTT NR RTT REO TT OT 24 ee 25 Input Se OULD E 25 WIOIKENEQUIA 0 A PRE AAA ENEE EE AE EEN 25 Referentes eer Geen e ee
16. cations allow to mark PCM elements e g an operation of a component to be monitored during analysis using a metric such as response time Service level objectives specify thresholds for these metrics allowing software architects to manifest their quality related requirements Self adaptation rules can react on changes of monitored values For example when a certain response time threshold is exceeded an adaptation rule could trigger a scaling out of bottleneck components ll Input amp Output Creating Extended PCM instances is mainly based on the input of the software architects that create these instances Software architects can however be supported by using Architectural Templates see Architectural Templates ATs or by reverse engineering partial Extended PCM instances from source code see Extractor The main output for specifying an Extended PCM is the Extended PCM instance itself It can be used for documenting a system s architecture and is part of a ScaleDL instance the Analyser input see Analyser l Minimal Example To create an Extended PCM instance software architects conduct the following actions 1 Switch to the Analyser Perspective second button of the image below E O OA Ed ER JE 2 Create an Extended PCM instance by following the Palladio Workshop white paper and the Analyser screencast series For example PCM component repositories are created via the Ka button shown in the image above
17. e project selected It also exposes metrics used in extraction o Results all extraction data are persisted in results folder and showed through this component extracted PCM models modisco models e Static Spotter o Input Source Decorator model produced by SoMoX with corresponding PCM models o Run view and manipulation with Static Spotter catalog Run static analysis o Results persists and displays all anti patterns found in the results e Dynamic Spotter o Server configuration of Spotter server In CSE user can work with integrated one or he can configure address of external spotter server Input Provides Instrumentations and Measurements configurations o Run Provides basic Dynamic Spotter configurations with Workflow and Hierarchy selection o Results all results are persisted in results folder and showed through this component l Workflow The workflow diagram represents all possible paths that user can take in the CloudScale Environment to analyse scalability of the system Figure 2 Workflow diagram with successfully validated alternatvies The diagram itself is composed of 4 main groups Analyser Extractor Spotter and ScaleDL Each group defines a set of states and how they are depended on each other while external boxes represents actions i e generate import use that can be executed to connect results of one tool to the input of another The tools groups contain three states
18. elow Creating the PCM model The first step is to create or import a Palladio Component Model PCM of the system to evolve including a Usage model with at least one Usage Scenario This task can e g be initialised from the Analyser tab of the Dashboard where the set of PCM models can be created or imported by selecting Input pressing the Create new button and selecting one of the input alternatives See the Analyser and PCM chapter for details on this Creating LIMBO models The next step is to create LIMBO model of how the load and selected work parameters from the PCM model evolve over time LIMBO models can be created from the LIMBO tab of the Dashboard From the dialog that appears when selecting to create a LIMBO model it is possible to create either an empty LIMBO model a model based on a template or to select a more advanced setup through a wizard Note that the wizard interface cannot be used to edit the model after it has first been created M Dashboard X 12 pcm usagemo EI pcm reposit f pcm system 3 m Extractor Analyser Dynamic Spotter Static Spotter Overview Limbo Usage evolution arrival rate editor v Model Sequence Linear trend Resource limbo dlim Y TimeDependentFunctionContainer Container lt gt LinearTrend 0 0 29 97 arrival rates 0 0 time 600 1 Linear trend Alternative is valid E Properties 2 Validation status
19. em The usage model file is found under the subfolder of your selected input alternative for the analyser while the LIMBO models are found under the LIMBO subfolder of the ScaleDL models Once the files have been loaded make sure the Usage element in the editor is selected Click on the Load evolution property and select the sequence from the LIMBO model that appears in the list Then click on the Scenario property and select the usage scenario from Palladio that appears in the list The editor should now look similar to what is shown in the next figure Remember to save the model once the edits have been completed If there is any work parameters in your model that you want to describe evolution of add one Work Parameter Evolution childs to the Usage for each of them From the Evolution property select the LIMBO sequence that describe the work parameter evolution remember that each LIMBO model must first be first be imported into your model Set the Variable Characterisation property to point to the variable from the PCM model to evolve eoo CloudScale wi S Project Explorer 5 BS sr pg m Dashboard ImageArchiveAnalyser LO analyser usageevolution 5 210 E default system_diagram D Resource Set ka My resourceenvironment gt ES My resourceenvironment_diagram 3 prop alt gt oi publishOpen usagemodel gt zf publishOpen usagemodel_diagran gt SS Alternative_2015 07 03_02 45 35 gt SS Alternative_2015 07 03_02 50 03 gt Zi Con
20. evolution of a set of variables applies to a usage model expressed in the Palladio Component Model PCM The actual modelling of the evolution of a single variable e g arrival rates or population in load or average video size in work is done using the LIMBO Load Intensity Modelling Tool The details of using the Palladio and LIMBO are not covered here but we refer instead to the PCM chapter of this document the Usage Evolution screencast and to the tutorials and screencasts on the Palladio home pages Il Relation to Workflow An instance of the Usage Evolution is part of a ScaleDL instance and defined under the ScaleDL step in the workflow It is an input of the Analyzer see Workflow Input amp Output Walkthrough The following is an overview of the steps for creating and running a usage evolution simulation 1 Create a Palladio Component Model PCM of the system to evolve including a Usage model with at least one Usage Scenario 2 Create LIMBO models for how load and selected work parameters e g amount of data to process evolves time 3 Create a Usage Evolution model that maps how LIMBO models are used to describe the evolution of load and work parameters from a selected usage scenario 4 Create a run configuration 5 Run the simulation see details under the Analyser section 6 optional Adding additional usage and evolution models Each of the steps are described in more detail in the sub sections b
21. figuration O Results gt Dynamic Spotter gt gt Extractor gt Static Spotter Zi Generated models wC ScaleDL models SS Overview v Usage Evolution vi Alternative_2015 07 03_03 30 30 limbo dlim Selection Parent List Tree Table Tree with Columns Y oi platform resource ImageArchiveAnalyser 20Analyser Configuration Alternat Y Usage Evolution aName Y Usage aName Work Parameter Evolution gt oi platform resource ImageArchiveAnalyser 20Analyser 20Input Alternative_2 gt Be platform resource ImageArchiveAnalyser ScaleDL 20models Usage 20Evoluti gt LO platform resource ImageArchiveAnalyser 20Analyser 20Input Alternative_2 gt oi platform resource ImageArchiveAnalyser 20Analyser 20Input Alternative_2 gt oi pathmap PCM_MODELS PrimitiveTypes repository gt oi pathmap PCM_MODELS Palladio resourcetype 4 P Project workflow diagram ImageAr S E Properties 23 Validation 33 Arrival Ra 4 Simulatio E Console Ia Pale Oke gt Property Value Entity Name 1 aName ld _p_Tb8CGEEeWLbqKyHprDbw Load Evolution Sequence Sequence Scenario 4 Usage Scenario publishimageOper Running the simulation Open the Run tab under the Analyser in the Dashboard Select the configuration you want to run and press the Run button at the bottom right of the screen To see the result first check that you are in the Analyser perspective click the graph icon in the top left of the window The res
22. hat configurations are required and provides option to run the tool Since all integrated tools contain theirs own run configuration user interfaces the dashboard does not try to duplicate it however it tries to provide easier access to the configurations through pre configuring what is possible giving users options to create delete and modify configurations and run them Results section displays all results from tool runs Since the tools also provide their own user interfaces for displaying results the results component shows which results are available status of the run and provide an option to open specific result in dedicated component Nevertheless the dashboard is currently used only for integrated tools it also provides the extension point where additional plugins can extend its functionality by providing new tools and or operations e Analyser O Input Input models PCM with addition to automatically create new ones or import existing ones and apply Architectural Templates o Run convenient manipulation of the Experiments model and running capabilities supporting Scalability Capacity and Normal run configurations o Results persist entire result sets and shows most important informations including charts with possibility to open dedicated EDP2 views e Extractor o Input currently not visible since input can for now only be already imported Java projects o Run automatically configures Modisco and SoMoX targets based on th
23. ition for the Analyser it stops as soon as 100 measurements have been taken 4 Run the Analyser by pressing the Run button and wait for its completion the console output states Workflow engine completed task on completion 5 Investigate analysis results via the Results tab Clustering Composition Threshold Min End Value 25 Clustering Composition Threshold Max Start Value 100 Response Time 5 BEER EEE GE I References The CloudScale Method includes a dedicated step for running the Analyser and for investigating its analysis results Static Spotter Introduction The Static Spotter is a reverse engineering tool based on the Reclipse for automatic detection of so called search patterns which then are interpreted as a potential scalability anti patterns All scalability anti patterns are defined in the pattern catalogue Later on the Static Spotter is searching for the anti patterns according to the pattern catalogue ll Relation to workflow r Extractor Analyser e Spotter wW a Results SE Figure 4 StaticSpotter workflow l Problem A software system suffers scalability or performance issues while being development without consideration of scalability or performa
24. nalyze ScaleDL models regarding scalability elasticity and efficiency of cloud computing applications at design time For these capabilities CloudScale integrated novel metrics for such properties into Analyzer Analyses are based on analytical solvers and or or simulations Analyzer particularly supports to analyze self adaptive systems e g systems that can dynamically scale out and in ll Relation to Workflow The Analyser is important for the right hand side of the Workflow see image below It is capable of providing scalability elasticity and cost efficiency analysis results based on a fully specified ScaleDL instance Such analysis results can already be provided before the system is accordingly implemented e g at early design time or to investigate modification scenarios r Extractor P Analyser uu aC Static spotter Dynamic spotter Results Results B e Spotter ee oe PU Figure 3 Analyser workflow l Problem Scalability elasticity and cost efficiency problems are often detected too late i e during system testing and or system operation This approach leads to the risk of SLO violating systems and expensive re implementations Moreover different design variants remain uninvestigated because it would be too extensive to implement and test all potential design variants ll Features CloudScale s Analyser measures several scalabilit
25. nce This would become even worse when being migrated to the Cloud computing environment For example the system gets and pays more resources from the Cloud providers but does not gain in its performance l Features The Static Spotter exams the ScaleDL model It uses predefined pattern or anti pattern catalogues to detect patterns and anti patterns automatically l Input amp Output The Static Spotter takes ScaleDL models as input After searching for predefined patterns or anti patterns it generates a list of pattern or anti pattern candidates The candidates could be methods classes or components Walkthrough 1 Goto Static Spotter Tab click input gt create 2 Select Import Extractor result 3 Click Run gt Create Select Input BROW OSE F Be wwe veces Q 300 Get tw H BR O TT Co Carrey there aro no netas 1o play Cora ama one NH 9 ProaucerConumar sce Sime Spotter par Import options Sech ono ol Pa porta options eer peje port Extractor SS Drees naw input alero rom enterrar sourcecode decorator model Resource internal_architecture_model sourcecode gt l Model Repository aName Resource internal_architecture_model repository System Resource internal_architecture_mode Input alternative Catalog and Engine models gt il Model PSCatalog New Catal
26. og Resource catalog pes gt 8 Model EPackage annotations Resource catalog pos ecore e success see resun Run 4 Run A D a gt E Model PSCatalog New Catalog Resource catalog pos if Model EPackage annotations Resource catalog pes ecore 5 Goto Results under Main menu and see the results Dynamic Spotter Introduction Dynamic Spotter is a framework for measurement based automatic detection of software performance problems in Java based enterprise software systems It combines the concepts of software performance anti patterns with systematic experimentation l Reference to workflow r Extractor mm Analyser po x E Figure 5 StaticSpotter workflow l Problem A software system suffers scalability or performance issues while being development without consideration of scalability or performance This would become even worse when being migrated to the Cloud computing environment Although Static Spotter is able to find potential scalability or performance anti patterns it is hard to tell which one is affecting actually ll Features Dynamic Spotter uses systematic experimentation analyses measurement and detects the actual affecting anti patterns in the system Walkthrough 1 In order to run Dynamic Spotter a running server is needed Go to
27. rresponding diagrams Additionally all PCM models imported into an Overview model are stored in the imported subfolder For each integrated tool corresponding plugin creates folder where it stores all tool specific information In general all tools needs input configuration and results therefore these are also present in first level in subfolder Further down it depends on specific integration which data is stored and how the data is organized Nevertheless users can always find all tool s specific information e g models configurations results etc inside these folders l Dashboard To provide unified access to all the integrated tools project Dashboard component has been developed In general each integrated tool needs some input run configuration and at the end it provides some results Taking this into account dashboard contains main user interface component for each tool which is further divided into three sections input configuration results Input section defines what the input is e g complete PCM model for Analyser and how to acquire it e g output of the Overview transformation is input into the Analyser Supporting different CloudScale Method paths one can use the tool independently i e define manual input or connect other tool s output to its input In the current state the integration mechanism does not yet support alternative inputs which is planned for the next year Configuration section defines w
28. rspective of the tools actions and views are hidden since all functionality is available through integration components To use Tools in a stand alone fashion see Tools perspectives section l CloudScale Environment Project The initial step in using the CloudScale Environment is creation of a project A project represents a system being analyzed by the tool user As showed in Figure 1 a project initially contains project specific files dedicated ScaleDL models folder with automatically generated ScaleDL Overview model and diagram and dedicated tool s folders 5 CloudStore Environment CloudScale Project Example R Analyser 4 Input e Alternative 2015 05 14 05 35 16 e Alternative 2015 05 20 01 48 42 XX Configuration Results Dynamic Spotter Extractor Static Spotter 3 Generated models Cg ScaleDL models ze method workflow A project cse CS README md Figure 1 CloudScale Environment project Project specific files are project cse and method workflow First contains general information about the project location of files results etc and is presented through the Project Dashboard The method workflow on the other hand stores the current state of the project models validations enabled and disabled steps etc which are presented through the workflow diagram ScaleDL models folder contains as name suggests all ScaleDL models Overview Usage Evolution and Architectural Templates and co
29. s gt Show View action Alt Shift Q Q The CloudScale Environment is designed to be fully extensible giving the possibility to the open source community of improving and adapting it to its own needs All the integrated tools Palladio SoMox Reclipse and the SoPeCo will be available from within the application through separate perspectives however the main perspective CloudScale perspective will try to cover as much functionality as needed through dashboard and workflow components see 4 2 to achieve a seamless and integrated user experience providing functionalities described before e Analyser Palladio Palladio is a software architecture simulation framework which analyses software at the model level for performance bottlenecks scalability issues and reliability threats and allows for a subsequent optimisation In the CloudScale Environment mainly the backend engines will be used to analyse the user s application This will be achieved by automatically transforming ScaleDL Overview models into the Palladio Component Models structure on which Palladio operates e Extractor SoMoX SoMoX provides clustering based architecture reconstruction to recover the architecture of a software system from source code The clustering mechanism extracts a software architecture based on source code metrics and construct PCM models to be used by Analyser e Dynamic and Static Spotter Spotter is a tool for identifying spotting scalability
30. t Ee Seege 26 IN 27 a EE 27 Relation to WOTKPIOW get eebe A AA A AEN ate 27 POD sarita it at E A IA EA A A RARA 27 HOLD A iia A id aba 28 ele He OWED EE 28 WOIKEALOUQK te es ad a ea o e e dE 28 Referentes aurca EE Ee ee 29 Static DA di lada 30 Introduction EE 30 Relation TO WO de 30 PODIO daria AAA AAA AAA AA A AAA ARA 30 ee 30 PUE DU a a AR EA E 31 WARO g o EI to do al o e Df ed 31 DYNAMIC PA da Ea REESE 33 A E 33 Referente to MIER enee cave EES EENS Problem Features Walkthrough Installation Requirements e Java 7 JRE is not yet included e Windows OSX or Linux operating system 32 or 64 bits o If 32 bit Java is used on 64 bit OS 32 bit bundle needs to be downloaded l Download o CSE bundles available at http www cloudscale project eu results tools o OSis automatically detected o Choose between Release and Nightly version l Installation amp Run 1 Download bundle 2 Extract Unzip bundle 3 Open folder and run Environment eclipse Development l Maven Tycho 1 Clone the repository o S git clone https github com CloudScale Project Environment git 2 Build Cloudscale Environment o S mvn package 3 Run Linux MacOS Windows distribution o Bundle location plugins eu cloudscaleproject env master target products Eclipse IDE 1 Download and install Eclipse Luna for RCP and RAP 2 Download and install Eclipse plugin dependencies for CloudScale development
31. ults of the simulation appears in the Experiments view that is shown in the tab next to the Project Explorer in the Analyser perspective See the Analyser section for more details on running simulations Adding additional usage and evolution models In some cases it is will be useful to create more than one usage scenario and or usage evolution description to analyse different uses of the systems or how it will evolve in different future scenarios To add this from the Input tab for the Analyser tab in the dashboard e Click the create button under the alternative inputs column to create an additional analyser input alternative In the wizard that appears import the models with exception of the usage model from the initial alternative to them e Create a new usage model by clicking the Create button on the right hand side of the Analyser tab and select Usage Model and Usage Evolution Model e Adda new run configuration where the Usage Evolution for the alternative will be located e Add content to the new models e As required import or create new Limbo models to describe the evolution of individual parameters and refer to these from the usage evolution mode as described in the previous section Introduction The Extractor is a reverse engineering tool for automatic model extraction It parses source code and generates partial ScaleDL models that are further used by the Analyser ll Relation to Workflow
32. y elasticity and cost efficiency metrics based on a ScaleDL instance as an input The metrics of interest can be configured by software architects l Input amp Output As an input Analyser takes a ScaleDL instance Extended PCM Usage Evolution Architectural Templates ScaleDL Overview and is configuration via an Experiment Automation instance The latter is a dedicated configuration model for the Analyser e g allowing to configure metric measurements and how often and how long Analyser runs are conducted The output of the Analyser are analysis results The CloudScale Environment can store load and visualize these results Walkthrough 1 Switch to the CloudScale Perspective first button of the image below Su I Dashboard Minimal Environment 33 2 Load your input ScaleDL instance into od Ainme Emir t he Ana lyser by using t he u Create Sp Extractor Analyser Ge Spotter Static Spotter Overview Usage evolution wizard in the Input tab You can ee selection pas choose between different input sources for your ScaleDL instance as shown in the image below Create new input alternative New Analyser input pi import Extractor result Create new input alternative with empty PCM models Transfrom Overview model Alternative is valid 3 Configure an Analyser Run via the Run tab and configure Analyser parameters For example the image below sets a stop cond
Download Pdf Manuals
Related Search