ServiceInsight and ServicePulse retrieve data from a ServiceControl instance using an HTTP API. In some installations, that data may reside in multiple ServiceControl instances. The ServiceControl Remotes features allows a ServiceControl instance to aggregate data from other ServiceControl instances, providing a unified experience in ServiceInsight and ServicePulse.
Overview
One ServiceControl Error instance is designated as the primary instance. All other ServiceControl instances are remote instances. The HTTP API of the primary instance aggregates data from the primary instance and from all the remote instances. ServiceInsight and ServicePulse are configured to connect to the primary instance.
The term remote refers to the fact that remote instances are run in separate processes. The primary instance and one or more remote instances can run on the same machine.
In ServiceControl version 4 and later, a ServiceControl Error instance can be configured with remote instances that are also ServiceControl Error instances or are ServiceControl Audit instances. ServiceControl Audit instances cannot be configured as primary instances.
Default deployment
In ServiceControl version 4 and above, the ServiceControl Management utility creates a primary ServiceControl Error instance and a remote ServiceControl Audit instance.
graph TD
Endpoints[endpoints] -- send audits to --> AuditQueue(audit queue)
Endpoints -- send errors to --> ErrorQueue(error queue)
ErrorQueue -- ingested by --> ServiceControl[ServiceControl<br/>primary]
ServiceInsight -. connected to .-> ServiceControl
ServicePulse -. connected to .-> ServiceControl
AuditQueue -- ingested by --> ServiceControlAudit[ServiceControl Audit<br/>remote]
ServiceControl -. connected to .-> ServiceControlAudit
classDef Endpoints fill:#00A3C4,stroke:#00729C,color:#FFFFFF
classDef ServiceInsight fill:#878CAA,stroke:#585D80,color:#FFFFFF
classDef ServicePulse fill:#409393,stroke:#205B5D,color:#FFFFFF
classDef ServiceControlPrimary fill:#A84198,stroke:#92117E,color:#FFFFFF,stroke-width:4px
classDef ServiceControlRemote fill:#A84198,stroke:#92117E,color:#FFFFFF
class Endpoints Endpoints
class ServiceInsight ServiceInsight
class ServicePulse ServicePulse
class ServiceControl ServiceControlPrimary
class ServiceControlAudit ServiceControlRemote
Sharding audit messages with competing consumers
Two ServiceControl Audit instances ingest messages from the same audit queue. This approach can be used to scale out the ingestion of messages from high-volume audit queues.
graph TD
Endpoints[endpoints] -- send audits to --> AuditQueue(audit queue)
Endpoints -- send errors to --> ErrorQueue(error queue)
ErrorQueue -- ingested by --> ServiceControl[ServiceControl<br/>primary]
ServiceInsight -. connected to .-> ServiceControl
ServicePulse -. connected to .-> ServiceControl
AuditQueue -- ingested by --> ServiceControlAudit1[ServiceControl Audit 1<br/>remote]
ServiceControl -. connected to .-> ServiceControlAudit1
AuditQueue -- ingested by --> ServiceControlAudit2[ServiceControl Audit 2<br/>remote]
ServiceControl -. connected to .-> ServiceControlAudit2
classDef Endpoints fill:#00A3C4,stroke:#00729C,color:#FFFFFF
classDef ServiceInsight fill:#878CAA,stroke:#585D80,color:#FFFFFF
classDef ServicePulse fill:#409393,stroke:#205B5D,color:#FFFFFF
classDef ServiceControlPrimary fill:#A84198,stroke:#92117E,color:#FFFFFF,stroke-width:4px
classDef ServiceControlRemote fill:#A84198,stroke:#92117E,color:#FFFFFF
class Endpoints Endpoints
class ServiceInsight ServiceInsight
class ServicePulse ServicePulse
class ServiceControl ServiceControlPrimary
class ServiceControlAudit1,ServiceControlAudit2 ServiceControlRemote
Sharding audit messages with split audit queues
Endpoints are partitioned into groups. Each group sends messages to its own a different audit queue. Each audit queue is managed by a different ServiceControl Audit instance. This approach is useful if different audit retention periods are required for specific groups of endpoints.
graph TD
Endpoints1[endpoints<br/>group 1] -- send audits to --> AuditQueue1(audit queue 1)
Endpoints1 -- send errors to --> ErrorQueue(error queue)
Endpoints2[endpoints<br/>group 2] -- send errors to --> ErrorQueue
ErrorQueue -- ingested by --> ServiceControl[ServiceControl<br/>primary]
ServiceInsight -. connected to .->ServiceControl
ServicePulse -. connected to .-> ServiceControl
Endpoints2 -- send audits to--> AuditQueue2(audit queue 2)
AuditQueue1 -- ingested by --> ServiceControlAudit1[ServiceControl Audit 1<br/>remote]
ServiceControl -. connected to .-> ServiceControlAudit1
ServiceControl -. connected to .-> ServiceControlAudit2[ServiceControl Audit 2<br/>remote]
AuditQueue2 -- ingested by --> ServiceControlAudit2
classDef Endpoints fill:#00A3C4,stroke:#00729C,color:#FFFFFF
classDef ServiceInsight fill:#878CAA,stroke:#585D80,color:#FFFFFF
classDef ServicePulse fill:#409393,stroke:#205B5D,color:#FFFFFF
classDef ServiceControlPrimary fill:#A84198,stroke:#92117E,color:#FFFFFF,stroke-width:4px
classDef ServiceControlRemote fill:#A84198,stroke:#92117E,color:#FFFFFF
class Endpoints1,Endpoints2 Endpoints
class ServiceInsight ServiceInsight
class ServicePulse ServicePulse
class ServiceControl ServiceControlPrimary
class ServiceControlAudit1,ServiceControlAudit2 ServiceControlRemote
Multi-transport deployments
When a system uses multiple transports, the Messaging Bridge can be used to allow management of the entire system by single instances of ServicePulse and ServiceInsight.
graph TD
primaryA -. connected to .-> auditA
primaryA -. connected to .-> auditB
primaryA -. connected to .-> auditC
subgraph Transport A
ServicePulse[ServicePulse] -. connected to .-> primaryA
ServiceInsight[ServiceInsight] -. connected to .-> primaryA
endpointsA[endpoints] -- send errors to --> errorQueueA[error queue]
endpointsA -- send audits to --> auditQueueA[audit queue]
errorQueueA -- ingested by --> primaryA[ServiceControl<br/>primary]
auditQueueA -- ingested by --> auditA[ServiceControl Audit<br/>remote]
end
subgraph Transport B
endpointsB[endpoints] -- send errors to --> errorQueueB[error queue]
endpointsB -- send audits to --> auditQueueB[audit queue]
errorQueueB -- ingested by --> bridgeB[Bridge]
auditQueueB -- ingested by --> auditB[ServiceControl Audit<br/>remote]
bridgeB -- forwards --> errorQueueA
end
subgraph Transport C
endpointsC[endpoints] -- send errors to --> errorQueueC[error queue]
endpointsC -- send audits to --> auditQueueC[audit queue]
errorQueueC -- ingested by --> bridgeC[Bridge]
auditQueueC -- ingested by --> auditC[ServiceControl Audit<br/>remote]
bridgeC -- forwards --> errorQueueA
end
classDef Endpoints fill:#00A3C4,stroke:#00729C,color:#FFFFFF
classDef Bridge fill:#a8a032,stroke:#00729C,color:#FFFFFF
classDef ServiceInsight fill:#878CAA,stroke:#585D80,color:#FFFFFF
classDef ServicePulse fill:#409393,stroke:#205B5D,color:#FFFFFF
classDef ServiceControlPrimary fill:#A84198,stroke:#92117E,color:#FFFFFF,stroke-width:4px
classDef ServiceControlRemote fill:#A84198,stroke:#92117E,color:#FFFFFF
class endpointsA,endpointsB,endpointsC Endpoints
class bridgeB,bridgeC Bridge
class ServiceInsight ServiceInsight
class ServicePulse ServicePulse
class primaryA,primaryB ServiceControlPrimary
class auditA,auditB,auditC ServiceControlRemote
Multi-region deployments
It is possible to create a multi-region deployment using remotes.
graph TD
insight[ServiceInsight] -..-> crossRegionPrimary[Cross Region<br/>ServiceControl<br/>primary]
crossRegionPrimary -. connected to .-> primaryB
subgraph Region B
primaryB -. connected to .-> auditB
servicePulseB[ServicePulse] -. connected to .-> primaryB
endpointsB[endpoints] -- send errors to --> errorQueueB[error queue]
endpointsB -- send audits to --> auditQueueB[audit queue]
errorQueueB -- ingested by --> primaryB[ServiceControl<br/>primary]
auditQueueB -- ingested by --> auditB[ServiceControl Audit<br/>remote]
end
subgraph Region A
endpointsA[endpoints] -- send errors to --> errorQueueA[error queue]
endpointsA -- send audits to --> auditQueueA[audit queue]
errorQueueA -- ingested by --> primaryA[ServiceControl<br/>primary]
auditQueueA -- ingested by --> auditA[ServiceControl Audit<br/>remote]
primaryA -. connected to .-> auditA
servicePulseA[ServicePulse] -. connected to .-> primaryA
end
crossRegionPrimary -. connected to .-> primaryA
classDef Endpoints fill:#00A3C4,stroke:#00729C,color:#FFFFFF
classDef ServiceInsight fill:#878CAA,stroke:#585D80,color:#FFFFFF
classDef ServicePulse fill:#409393,stroke:#205B5D,color:#FFFFFF
classDef ServiceControlPrimary fill:#A84198,stroke:#92117E,color:#FFFFFF,stroke-width:4px
classDef ServiceControlRemote fill:#A84198,stroke:#92117E,color:#FFFFFF
class endpointsA,endpointsB Endpoints
class insight ServiceInsight
class servicePulseA,servicePulseB ServicePulse
class primaryA,primaryB,crossRegionPrimary ServiceControlPrimary
class auditA,auditB ServiceControlRemote
In this deployment, each region has a full ServiceControl installation with a primary Error instance and an Audit instance. Each region can be managed and controlled via a dedicated ServicePulse.
A new cross-region primary instance is added to allow ServiceInsight to show messages from both regions. This cross-region instance includes each region-specific primary instance as a remote allowing it to query messages from both. The cross-region instance must disable error message ingestion management by setting with the value ServiceControl/IngestErrorMessages
option to false
.
Zero downtime upgrades
The remotes feature can be used to perform zero downtime upgrades of Audit instances.
Configuration
Remote instances are listed in the ServiceControl/RemoteInstances
app setting in the primary instance configuration file. The value of this setting is a JSON array of remote instances. Each entry requires an api_url
property specifying the API URL of the remote instance. For ServiceControl version 3 and earlier, each entry requires a queue_address
property specifying the queue address of the remote instance.
Changes to the configuration file do not take effect until the primary instance is restarted.
Version 4 and later
<configuration>
<appSettings>
<add key="ServiceControl/RemoteInstances" value="[{"api_uri":"http://localhost:33334/api"}]"/>
</appSettings>/
</configuration>
Version 3 and earlier
<configuration>
<appSettings>
<add key="ServiceControl/RemoteInstances" value="[{"api_uri":"http://localhost:33334/api", "queue_address":"Particular.ServiceControl.Remote"}]"/>
</appSettings>/
</configuration>
Managing remote instances using PowerShell
The following cmdlets are available in ServiceControl version 4 and above, for the management of remote instances:
Alias | Cmdlet |
---|
sc-addremote | Add-ServiceControlRemote |
sc-deleteremote | Remove-ServiceControlRemote |
sc-remotes | Get-ServiceControlRemotes |
The names and addresses of instances are controlled by the cmdlets for managing ServiceControl Error and Audit instances.
Add a remote instance
Add-ServiceControlRemote
adds a remote instance to a primary instance.
Add-ServiceControlRemote `
-Name $serviceControlInstance.Name `
-RemoteInstanceAddress $auditInstance.Url
Remove a remote instance
Remove-ServiceControlRemote
removes a remote instance from a primary instance.
Remove-ServiceControlRemote `
-Name Test.ServiceControl `
-RemoteInstanceAddress http://localhost:44444/api
List remote instances
Get-ServiceControlRemotes
gets a list of remote instances from a primary instance.
Get-ServiceControlRemotes -Name Particular.ServiceControl
Changing the address of a remote instance
To change the address of a remote instance to a new host and/or port number:
- Remove the current address from the list of remote instances:
Remove-ServiceControlRemote -Name $primaryServiceControl.Name -RemoteInstanceAddress $currentAddress
- Restart the primary instance to refresh the list of remote instances
- Stop the remote instance
- Change the host and/or port number of the remote instance using the ServiceControl Management utility
- Start the remote instance at its new address
- Add the new address to the list of remote instances:
Add-ServiceControlRemote -Name $primaryServiceControl.Name -RemoteInstanceAddress $newAddress
- Restart the primary instance to refresh the list of remote instances
Considerations
- Pagination in ServiceInsight may not work as expected. For example, each page may contain a different number of items, depending on how those items are distributed across the various ServiceControl instances.
- If the primary instance cannot contact a given remote instance, data from that remote instance will not be included in any views in either ServiceInsight or ServicePulse.
- Multi-instance configuration is not possible the ServiceControl Management utility.
- Incorrect configuration may cause cyclical dependencies. For example, instance A may attempt to get data from instance B, and instance B may attempt to get data from instance A.
- It is recommended to run only one primary instance. Multiple primary instances are not recommended.