Shared memory overview

Shared memory is a mechanism for sharing commonly used data across multiple Teamcenter processes. The shared memory implementation is applied to all the collocated Teamcenter processes that are run on the same machine, and the shared memory appears in the address space of all Teamcenter servers, utility processes, and service (daemon) processes on the machine. When large amounts of data need to be shared, shared memory is the fastest interprocess communication mechanism between co-located processes. After the memory is mapped into the address space of the processes that are sharing the memory region, the processes can access the shared memory area like regular working memory with no overhead incurred. Because only one copy of the data in the shared memory is resident for all the processes that are sharing the memory region, using shared memory for common data reduces the overall system memory footprint. Shared memory provides the following benefits: Reduces the memory footprint for each server process and utility process. Large amounts of data are loaded into shared memory. This data includes text server data, site preference data, localized list of values data, and metadata. As a result, the private memory footprint for a server process is reduced. For example, in a Teamcenter 9.1.2 environment, the savings in the private memory footprint for a server process are as follows: Text server: 6MB Preferences: 2–4 MB Lists of values: 5 MB Metadata: 19MB Improves Teamcenter scalability Because the private memory footprint is reduced per server, more Teamcenter servers can be configured on the same machine than previously. Improves Teamcenter performance All the shared data is populated once in shared memory and saved in local backing store files. After backing store files are generated, all the subsequent servers do not need to populate data into shared memory. Therefore, the logon time for all the subsequent servers is improved. In the case of the shared metadata, the metadata cache is pregenerated (for example, during install and upgrade). This helps to reduce further the metadata population time. As a result, there is approximately 18% improvement in logon time. Shared metadata memory helps to improve overall performance for other use cases as well. Each server is able to access the already-populated shared metadata cache that would otherwise have to be on-demand initialized by each server if shared metadata cache were not used.