![Example: List Hosted Service[OperationContract(AsyncPattern = true)][WebGet(UriTemplate = @"{subscriptionId}/services/hostedservices")]IAsyncResult BeginListHostedServices(string subscriptionId, AsyncCallback callback, object state);HostedServiceList EndListHostedServices(IAsyncResult asyncResult);](https://image.slidesharecdn.com/vonneumannworkers-110516093202-phpapp02/85/Von-neumann-workers-29-320.jpg)
Von neumann workers
- 2. Von NeumannWorkersRiccardo Becker| Practice Manager Microsoft| Logica
- 3. AgendaIntroduction (and turn onsomecloud stuff)
- 4. The Worker
- 6. Azure Service Management API
- 7. Elasticity
- 9. Summary & Next steps
- 10. Q & AIntroductionRiccardo Becker13 years of LogicaPractice ManagerStrong technical Microsoft backgroundwww.twitter.com/riccardobeckerhttp://nl.linkedin.com/in/riccardobeckerREMINDER: turn onmyTestDeploymenton the Portal!
- 11. Long running tasks that are asynchronous that user does not have to wait for.
- 12. To host application services that do not require a user interface.
- 13. Background services listening to a queue.
- 14. Running TCP based services.
- 16. But everything has a price!The Worker
- 17. Can handle different tasks (fromsimplecalculation up to long running)
- 19. Has flexibility of loading and unloading code
- 20. Is able to report it´s health
- 21. Has a queue thatdeliverswork
- 22. Loads assemblies from Blob Storage
- 23. Reports health and activities on Service BusThe GenericWorker
- 24. The GenericWorkerSimple demoQuartz.NETforschedulingUses queues, blobs and reflection
- 26. Windows Azure Service Management APIREST APISSL and mutually authenticated using X.509Needsubscriptionid (see portal)Operationsonstorage accounts, hosted services, deployments, certificates, affinitygroups (bothsync & asyncops)Activitieslikeon Management Portal (buildyourown)
- 27. Service Management APICreate, Swap & Delete DeploymentChangeDeploymentConfigUpdate Deployment StatusUpgrade DeploymentWalk Upgrade DomainRebootRoleInstanceReimageRoleInstanceList throughthosted services, deployments, instances, certificates, affinitygroupsEveryasyncoperation returns request-idUserequest-id in call to GetOperation Status forAsynchronousOperations (https://management.core.windows.net/<subscription-id>/operations/<request-id>)
- 28. Example: List Hosted Services
- 29. Example: List Hosted Service[OperationContract(AsyncPattern = true)][WebGet(UriTemplate = @"{subscriptionId}/services/hostedservices")]IAsyncResult BeginListHostedServices(string subscriptionId, AsyncCallback callback, object state);HostedServiceList EndListHostedServices(IAsyncResult asyncResult);
- 32. Storage
- 33. AffinityGroups
- 34. CertificatesWhat is elasticity?The ability to add/removebasedonloadTakethe right decisionbasedon the right metricsRaw data is notmeaningfulforuspeopleApplysmoothing to makeithumaninterpretable
- 35. WorkLoadPatternsOn offGrowing Fast On and Off Compute Compute InactivityPeriod Average UsageUsageAverageTime Time Successful services needs to grow/scale
- 36. Keeping up w/ growth is big IT challenge
- 37. Complex lead time for deployment
- 38. On & off workloads (e.g. batch job)
- 39. Over provisioned capacity is wasted
- 40. Time to market can be cumbersome Unpredictable Bursting Predictable Bursting Compute Compute Average Usage Average Usage Time Time Unexpected/unplanned peak in demand
- 41. Sudden spike impacts performance
- 42. Can’t over provision for extreme cases
- 43. Services with micro seasonality trends
- 44. Peaks due to periodic increased demand
- 45. IT complexity and wasted capacity Performance Counters// Add the performance counters (CPU and memory)diagConfig.PerformanceCounters.DataSources.Add(newPerformanceCounterConfiguration(){CounterSpecifier = @"\Processor(_Total)\% Processor Time",SampleRate = TimeSpan.FromSeconds(5)});// Copy the diagnostic metrics to TableStorage every minute fordiagConfig.PerformanceCounters.ScheduledTransferPeriod = TimeSpan.FromMinutes(1); // Set the LogLevel and perioddiagConfig.DiagnosticInfrastructureLogs.ScheduledTransferLogLevelFilter = LogLevel.Error;diagConfig.DiagnosticInfrastructureLogs.ScheduledTransferPeriod = TimeSpan.FromMinutes(1); // Start the diagnostic monitorDiagnosticMonitordiagnosticMonitor = DiagnosticMonitor.Start(csa, diagConfig);
- 46. Performance Counters cont’dEvery 5 seconds 1 sampleTransfer everyminute all data TableStorageTake the Average of e.g. last hour to decideScale up / down basedonoutcome
- 49. ExponentialMoving Average Smoothingexample + demo public static IEnumerable<WADPerformanceCountersTable> SimpleMovingAverage(IEnumerable<WADPerformanceCountersTable> source) { double total = 0; List<WADPerformanceCountersTable> target = new List<WADPerformanceCountersTable>(); int i = 0; foreach (WADPerformanceCountersTable pcdp in source) { i += 1; total += pcdp.CounterValue; WADPerformanceCountersTable entry = new WADPerformanceCountersTable(); entry.PartitionKey = pcdp.PartitionKey; entry.RowKey = Guid.NewGuid().ToString(); entry.CounterName = pcdp.CounterName; entry.DeploymentId = pcdp.DeploymentId; entry.Role = pcdp.Role; entry.RoleInstance = pcdp.RoleInstance + ".Smoothed"; entry.Timestamp = pcdp.Timestamp; entry.EventTickCount = pcdp.EventTickCount; entry.CounterValue = (total / i); target.Add(entry); } return target; }}DEMO Diagnostics Manager Cerebrata
- 50. ScalingManualorUseService Management APIModify service configurationIn this case, a workerdecides to scalenotsomeoverlord (probablybetter)
- 51. Scaling//take the average of the last hourvar averageMetrics = perfContext.CreateQuery<WADPerformanceCountersTable>("WADperformanceCountersTable"); var average = (from d in averageMetrics where (String.Compare(d.RoleInstance, RoleEnvironment.CurrentRoleInstance.Id + ".Smoothed") == 0) && (Convert.ToDateTime(d.EventTickCount) >= DateTime.UtcNow.AddHours(-1)) select d).ToList<WADPerformanceCountersTable>(); double averageCPU = average.Average(a => a.CounterValue); // get the deployment involvedDeployment deployment = (from p in this.hostedServiceList from c in p.Deployments from d in c.RoleInstanceList where (this.hostedServiceList != null) where (p.Deployments != null) && (c.RoleInstanceList != null) where (d.RoleName == RoleEnvironment.CurrentRoleInstance.Role.Name) select c).FirstOrDefault();
- 52. Scaling demoVisual Studio 2010
- 53. Windows Azure PortalA sea of parts
- 55. Manufacture a copy of itselfautonomously
- 57. Most effectiveway of large-scaleoperationsWhat is selfreplicationaccording to Von Neumann?
- 58. A sea of partsVon NeumannonAzureAzureBlobStorageAssembly AAssembly DAssembly GAssembly JAssembly PAssembly SAssembly MAssembly BAssembly EAssembly KAssembly HAssembly QAssembly NAssembly TAssembly CAssembly FAssembly LAssembly RAssembly IAssembly OAssembly UAssemblyLineQueueWorkerRoleWorkerRoleWorkerRoleWorkerRoleWorkerRoleUniversal ConstructorWorkerRoleElasticity agentAssemblylinewatcher
- 60. SummaryPay-per-use, usewhatyoupayMakeyourworkerssweat!Be efficient and scaleup/downLet them run for at leastanhour (youpay)Have funexperimenting!
- 61. Next stepsReport on service busSingleton in multi-instanceforElasticity AgentProper managementActivelyunload codeWill post oncodeplex?