110 likes | 206 Views
AppFabric in the Enterprise. Supporting Web Farms Features that help performance, availability & scale. AppFabric Web Farm Support. WebFarm is supported scenario, but has no UI tooling Scripting configuration to: push settings out to all machines in farm Push setting out to new nodes
E N D
AppFabric in the Enterprise • Supporting Web Farms • Features that help performance, availability & scale
AppFabric Web Farm Support • WebFarm is supported scenario, but has no UI tooling • Scripting configuration to: • push settings out to all machines in farm • Push setting out to new nodes • Get content out with WebDeploy • Package entire server • Individual IIS applications • Script deployment
Approach to Data Management • Manage Size of Monitoring & Persistence SQL database stores • DB Size & Retention Policy Options • Purge Tracking Event Data: older than date, circular buffer trim • Purge Persistence Data: cleanup after complete
Configuring Monitoring Purge Policy • Manual cleaning • Unused workflow services via SQL • On demand Clear-ASMonitoringSqlDatabase • Configure Archive DB & Purge Policy • Set-ASMonitoringSqlDatabaseArchiveConfiguration • Configure in monitoring db (no ui) • ASConfigurationPropertiesTable • APEnabled: true (1) • APThreshold: 1.5 GB (1500) • APTrimPercentage: 10% (10) • APMaxEventAge: 180 days (180)
AutoStart • Initializes the worker process before the first message is received Services hosted in AppFabric use message-based activation The AutoStart feature improves host responsiveness with minor resource consumption penalty Only available on Windows 7 & Server 2008 R2
Instance Restart - resilience you get for “free” Client Server A Server B Host B Host A WF Service Instance (Running) WF Service Instance (Running) Workflow Management Service Locked to Host A Lock Expires Locked to Host B
Workflow Management Service • Resiliency Scenarios • Scenario 1: Host crashes midway thru executing a workflow • Workflow Management Service looks for this • Schedules workflow to run on another host • Scenario 2: Workflow goes idle and persists waiting for a delay, then machine dies. Machine brought back up long after delay interval elapsed. • Workflow Management Service checks automatically for expired timers • Resumes workflows with expired timers • Control Scenarios • Scenario 3: A user issues a Suspend (or similar) command in the AppFabric UI or via the PowerShell cmdlets. • Workflow Management Service polls for queued commands • Executes the corresponding operation on the Workflow Control Endpoint
Instance Lock Retry Client A Client B Workflow Host B Workflow Host A Unlocked Locked
Configuring Lock Management Defines pattern for time interval (exponential or linear) between retries after failing due to instance lock exception. How long a host holds a lock. Controls the interval between successive polls of the instance store.
AppFabric Configuration • Instance Restart • Lock Retry • AutoStart
Lab 10 • AppFabric in the Enterprise