Glossary

Retired Content

This content and the technology described is outdated and is no longer being maintained. For more information, see Transient Fault Handling.

patterns & practices Developer Center

application throttling. The process of reducing an application's resource requirements, typically to maintain the performance of core functionality in response to changes in workload. Examples would be switching off nonessential features, or switching to a lightweight UI.

autoscaling. Using an automated mechanism to scale a Windows Azure™ technology platform application.

blob lease. A mechanism in Microsoft Azure that ensures that only a single client can access a blob.

burst. A sudden increase in the workload for an Azure application.

cool-down period (in the context of the Autoscaling Application Block). The period of time after a scaling action has taken place against your Azure application during which no further autoscaling activities should take place. This allows the application to settle down after the scaling operation and helps to reduce the number of scaling actions that are performed. You can configure the value of the cool-down period independently for scale-up and scale-down operations. The default value of the cool-down period is 20 minutes.

constraint rules. Rules that set explicit boundaries on the scaling process by defining the minimum and maximum number of instances permitted during a given time period. You should set the minimum value to ensure that you continue to meet your service level agreements (SLAs). You should set the maximum value in order to limit your costs. Constraint rules consist of the maximum and minimum instance count boundaries, a rank, and optionally a timetable that defines when the rule is in effect. If there is no timetable, the rule is always in effect.

cost optimization (in the context of the Autoscaling Application Block). A way to ensure that you make the best use of your running role instances by starting them early in the clock hour and stopping them late in the clock hour.

data point. An instantaneous metric value with an associated timestamp. The following table shows some example data points.

Metric

Value

Timestamp

CPU usage

83.7%

10:03:56

Unprocessed orders

2873

10:04:13

detection strategy (in the context of the Transient Fault Handling Application Block). A definition of the logic used to identify transient errors in a service.

elasticity. The ability of an application to automatically scale to meet changing workload requirements.

freeze period (in the context of the Autoscaling Application Block). The period immediately after a deployment operation or a change to your Azure application's service configuration during which Azure does not allow any additional configuration changes. The duration of the freeze period is determined by Azure and is not configurable. The duration of the freeze period is typically a couple of minutes, but may vary.

horizontal scalability (in the context of the Autoscaling Application Block). The ability of your application to be scaled by adding more role instances to your hosted service.

management API certificate (in the context of Windows Azure). A certificate used to secure Azure Management Service API calls.

metric. A parameter that is measured. Examples include performance counters such as CPU usage and free memory, and business-related metrics such as the number of unprocessed orders and the number of registered tenants in the application. Metrics may also be defined as the result of a calculation such as queue length per instance in the case where multiple role instances share a queue.

operand. Defines how to calculate the value for a metric that can be used in a reactive rule expression. For example, you can create a performance counter operand that monitors the CPU usage for a worker role and calculates the average value over 10 minutes. Operands are used by reactive rules.

rank. A property of a rule that is used by the block to resolve conflicts between rules. The higher the rank, the higher the priority.

reactive rules. Rules that react to varying loads on your application and trigger a scaling action when an aggregate value derived from a set of data points exceeds a certain threshold.

retry strategy (in the context of the Transient Fault Handling Application Block). A definition of how many retries to attempt and the interval between each retry.

retry policy (in the context of the Transient Fault Handling Application Block).****The combination of a retry strategy and a detection strategy.

role. A service definition to deploy your application code to Windows Azure. An Azure application may consist of many web and worker roles.

role instance. An instance of a web or worker role running in Windows Azure. An individual web or worker role may have multiple running instances in order to make the role more reliable or capable of handling larger workloads.

scale group. A way to define autoscaling rules that can act on multiple roles at once. Scale groups help to minimize the number of rules you need to create and manage. They can include roles in different hosted services.

service certificate (in the context of Windows Azure). A certificate that an application running in Azure can use to encrypt or decrypt data.

service information. Defines the aspects of your Azure application that are relevant to the Autoscaling Application Block. For example, the Autoscaling Application Block uses the service information file to know which roles are available for scaling, or which queues are available for monitoring.

stabilization (in the context of the Autoscaling Application Block). Damping the scaling operations to prevent unnecessary oscillations in the number of role instances as a consequence of the autoscaling operations.

target (in the context of the Autoscaling Application Block). Identifies a web or worker role type that can have multiple running instances and that can be scaled. Autoscaling actions can specify changes to the instance count of a target when the action is performed by an autoscaling rule. Targets usually refer to roles in a different hosted service from the hosted service where the Autoscaling Application Block is hosted.

timetable (in the context of the Autoscaling Application Block). Determines when a constraint rule is active. If a constraint rule does not have a timetable, the rule will always be on.

transient fault. An error that is due to some transient condition. For example, a transient fault can occur when you use a cloud-hosted service such as Azure storage or SQL Azure™ technology platform and you lose your connection as a result of temporary resource shortages. The result is an error condition, but when you retry the same command a short time later, it may succeed because the connection has been restored.

vertical scalability (in the context of the Autoscaling Application Block). The ability for your application to be scaled by increasing the size of a role instance by using more CPU cores and/or more memory.

WASABiCmdlets. A collection of Windows PowerShell® Cmdlets that you can use to manage the Autoscaling Application Block.

Next Topic | Previous Topic | Home

Last built: June 7, 2012