Supermap iServer Java 6R Cluster 2-in-process Cluster

Recently, a colleague tested the performance of Supermap iServer Java 6R. When asked about cluster configuration and the concept of In-process clusters, simply configure and test it by yourself. This is the second article of Supermap iServer Java 6R cluster research.

 

In Supermap iServer, the GIS functions provided by service providers are encapsulated into coarse-grained modules by service components and then published as different types of services. That is, after a service request arrives at the service component, the service component searches for the corresponding service provider, which is finally processed by the service provider.

A Supermap iServer cluster is a service provider that provides the same GIS functions in a process to form a logical cluster, which improves the efficiency of high concurrent access in a single server environment.

For example. sxwu workspace is released as a GIS service. To improve the efficiency of high concurrent access to the GIS service in a single server environment, you can use world at the service provider layer. sxwu workspace builds two service providers, such as ugcmapprovider-world and ugcMapProvider-World1, and then publishes these two service providers to the GIS service at the same time, this GIS service is more stable and efficient than common GIS services when processing high-concurrency access. (It can be configured as multiple providers for the UDB data source ).

Note that service providers of different data sources can form in-process clusters as long as they provide the same GIS function. For example, wmsmapprovider-World is a service provider built by remote WMS. It provides the same GIS function as ugcmapprovider-world, so you can add wmsmapprovider-world to the in-process cluster, this further improves the stability and efficiency of the GIS service during high-concurrency access. In addition, the service provider and the service provider set can form an in-process cluster, provided that the service provider and a service provider in the service provider set have the same GIS function.

Figure 1

 

We have discussed precautions for In-process clusters. The following configuration is a test environment, that is, how to configure in-process clusters and use them in the test.

 

1. Open the Service Manager and add multiple map service providers with the same GIS functions (for example, map1 and MAP2). For details about the configuration, refer to 2 and 3. Configure service provider map1

Figure 2

Configure service provider MAP2 for example 3

Figure 3

2. configure a service component after the configuration is complete. Configure 4

 

Figure 4

3. After the configuration is complete, you can directly browse the service instance mapcon/rest. You can view the service instance by taking a map at will, as shown in Figure 5.

Figure 5

Click Silverlight to view the image.

There are two solutions to check whether the cluster is used.

A) using the httpwatch request tracking tools, you can see 6 images taken from http: // localhost: 8090/iServer/output/cache, or http: // localhost: 8090/iServer/outputc/cache. B) You can directly browse the directory generated by the image to see that there are images under both directories, and the image generation time is consistent.

In-process clusters are finally configured in a simple way. In actual deployment, performance and deployment need to be considered comprehensively. The above deployment method increases the request concurrency pressure at the service component layer, all requests must go through the component layer. It would be best to split the image server, and the image is still in the same directory under the two directories. performance tests are required to verify the optimal performance, in addition, from the returned results of the request, all the returned image requests are evenly distributed from the two image servers. It seems that the service provider has simply evenly distributed the requests. In addition, it is worth considering the cache storage location for subsequent deployment. The above is a simple introduction and Testing Solution for In-process clusters. I hope you can correct it.