Configuring and managing distributed analysis service
|
Feedback
|
The Distributed analysis service uses distributed computing technology to perform spatial analysis and data processing on very large volume spatial datasets. iServer supports visual configuration of the distributed analysis service. The following content shows how to configure it.
- Configuring a distributed computing cluster
To configure the distributed anaysis service, first you need to configure a computing cluster since the implementation of the distributed analysis related functions provided by the distributed analysis service is based on calling the capability of a distributed computing cluster.
Currently, iServer supports three types of computing clustrs. The configuration steps are as follows:
- Using the iServer built-in Spark cluster
- Prepare the environment. For details, see: The environment preparation before using iServer built-in Spark.
- Start the Spark cluster on the local iServer. For details, see: Start iServer built-in Spark cluster service.
- Using a Spark cluster outside of iServer
- If you don't have a Spark cluster available, you can Build your own Spark distributed cluster on other computers.
- If you already have a Spark cluster available, you need to configure it in iServer to use it. For details, see: Using a Spark cluster outside the iServer.
- Using a Hadoop Yarn cluster outside of iServer
- Before using the Hadoop Yarn cluster, you need to do some configurations on the machine where the iServer locates. For details, see: environment configurations before iServer uses the Hadoop Yarn cluster.
- Using a Hadoop Yarn cluster which disables Kerberos authentication. For detailed configurations, see: Using a Hadoop Yarn cluster. If you need to set up a Hadoop Yarn cluster without Kerberos enabled, see: Building a Hadoop Yarn cluster without Kerberos authentication.
- Using a Hadoop Yarn cluster with Kerberos authentication enabled. For detailed configurations, see: Using a Hadoop Yarn cluster with Kerberos authentication. If you need to set up a Hadoop Yarn cluster with Kerberos authentication enabled, see: Building a Hadoop Yarn cluster with Kerberos authentication. Using a Hadoop Yarn cluster with Kerberos authentication enabled. For detailed configurations, see Using a Hadoop Yarn Cluster with Kerberos Authentication. If you need to set up a Hadoop Yarn cluster with Kerberos authentication enabled, see Building a Hadoop Yarn cluster with Kerberos authentication.
Note: iServer also provides you with some Spark cluster optimization configurations to improve service performance.
- Configure the settings of the distributed analysis service
Steps:
- After launching iServer on the primary node machine, enter the distributed analysis page.
- Click the "Clusters">"Distributed Analysis Services" tab in the iServer Service Manager (http://supermapiserver:8090/iserver/manager)
- Check "Enable service" to configure the following parameters:
- Settings of the associated iServer service
- Service address: The address of the iserver service used when the result of the distributed analysis is published as a iServer service. Optional parameter. Defaults to http://localhost:8090/iserver. (At present, only the associated iServer which locates on the same machine as the current iServer is supported.)
- Token: The token of the associated iServer service. Required parameter.
- Result output setting: Supports output to local disk, iServer DataStore, registered database (HBase, PostGIS, and PGSQL). Default: local disk.
- Whether publish the analysis result as a service. Default: The results of the analysis are published as a service.
- Check (or default): Support for publishing analysis results as rest map service, rest data service. Requires to config the associated iServer service. The specific configuration is as follows:
- Associated iServer service address: The address of the iserver service used when the result of the distributed analysis is published as a iServer service. Optional parameter. Defaults to http://localhost:8090/iserver. (At present, only the associated iServer which locates on the same machine as the current iServer is supported.)
- Associated service token: The token of the associated iServer service. Required parameter.
- Not checked.The results of the analysis will not be published as an iServer service.
- JVM maximum heap size: jvm maximum heap memory. Optional parameter. Default: 1024m. (If the analysis result data is relatively large, it is recommended to modify this parameter to 4g or more)
- Process port: The port of the distributed analytics service process. Optional parameter. Default: 6675
- Click "Save".
- Configure the security settings of the distributed analysis service
- iServer default: Users of the ADMIN role can create distributed analysis jobs.
- If users of other roles want to access and use the service, you need to authorize it. The configuration steps are as follows:
-
- After starting iServer on the primary machine, log in as the system administrator.
- Visit the Distributed Analysis Management page ( http://supermapiserver:8090/iserver/manager/services/distributedanalyst)
- Check "distributedanalyst/rest" and click "Authorization Service" to authorize permissions for users to access this service instance according to your needs.
- Enter a specific distributed analysis job creating page.
After enable the distributed analytics servic, you can add rich visual rendering effects to the published services via Setting thematic map template.