__cognos Development of Cognos principle

Cognos 8.4 Official documents The main process of http://www.itpub.net/thread-1287663-1-1.html Cognos query is: 1, select query content (report or data column) 2, enter run parameters (query conditions) 3, syntax mapping, Translate into specific process (execution plan) 4, Business database run Phase 5, Cognos local temporary file generation phase. 6, the result data generation Stage 7, the data returns the stage above 7 processes for the Cognos query execution main process, is affects the query performance the main link, its actual execution link needs more, for instance the server assignment, the result data return asynchronous polling and so on.   The execution route labeled red numbers in the image above is frequently problematic, as explained below: 1. The reason for this step is that it may be possible to retrieve too many numbers from the database for the user to choose from, and the second is to do too much complex processing in the client (ie browser). Three is a large number of user-selected data submitted to the Cognos application server. This place is going to be a bottleneck because the optional data for a client is routed from the data layer through the application layer through the Network Web protocol to the browser client; B Browser client's ability to data processing is weak; The results of the C selection go through the network to the Web layer and then to the application layer. and the network transmission spans the WAN link. 4, mainly a database in the processing process may receive a Cognos syntax parser generated by inefficient Sql;b database generated a large number of data records returned to the application server (more than 10,000); C Cognos products themselves in the process of receiving data using inefficient algorithms (single-threaded processing, It is possible that the IO optimization policy, such as file system write caching, is not included. 5, is the process of generating the results file, in the process of Cognos algorithm problems, inefficient operation, the general generation of 20M data per minute, and only single-threaded processing. This process is due to the final requirements, although it is the key to performance, but it can be resolved at the design stage through technical means, so the issue will be discussed later. 6, the process is the result of data generation through the network back to the client process, because the process is generally transmitted through the WAN, the available bandwidth in the general 512k, the amount of data per minute may be 3M, even in accordance with the 1M bandwidth, the amount of data can be transmitted per minute can only 6M, If there is a data volume of 20M data, the transmission process will take 3-4 minutes.        by detailing the details of the Cognos query, we can see that in Process 4, the SQL execution process and the temporary files (several g are usually common) build processes are the easiest part of our design phase to avoid performance problems, Because other links are either subject toSeek, or be subject to infrastructure (e.g. network). As a result, we will give a detailed description of the strategy for optimizing performance through the design process.