This chapter describes how to use Express clusters to provide services to data warehouse clusters, create and query tables, and check whether a data warehouse cluster is accelerated.
- Currently, the Express feature is in the OBT. You can apply for it. Move the cursor to Service Tickets in the upper right corner of the management console and click Create Service Ticket to submit the application for enabling the OBT feature.
Express clusters are independent of data warehouse clusters but are located at the same level as the data warehouse clusters. An Express cluster provides computing acceleration services for one or more data warehouse clusters. By default, Express clusters provide services for all data warehouse clusters of the matched version.
During the actual use of data warehouse clusters, the aged service data is put into OBS for joint queries with internal GaussDB(DWS) tables. Foreign table querying of GaussDB(DWS) uses the full-table scanning mode. The processing speed depends on the cluster performance. If a data warehouse cluster has a small number of nodes and low specifications, the processing speed is slow.
The Express cluster is a shared cluster that speeds up the query on OBS foreign tables. When you perform OBS foreign table querying in a data warehouse cluster, the data scanning operator, filtering operator, and aggregation operator are pushed to the Express cluster for calculation. The Express cluster has a larger scale and higher performance than the user cluster, so a better acceleration effect can be achieved.
You do not need to modify existing services when using the Express cluster. When you perform OBS foreign table querying, the system automatically determines whether to execute Express query acceleration based on the resource usage. If the acceleration is required, the execution plan of operators such as data scanning, filtering, and aggregation is pushed to the Express cluster. After the calculation is complete, the Express cluster returns the data to the data warehouse cluster, which performs subsequent queries based on the execution plan.