Virtual Clusters GBASE can contain one or more VCs in a universal virtual cluster. Each VC is a physical cluster, each VC is managed by the same set of Coordinator Clusters, and each VC runs independently within the scope of the virtual cluster and shares a unified entrance. Each virtual cluster (VC) consists of a set of Data nodes, and all virtual clusters are controlled by a set of Coordinator Clusters. Each virtual cluster operates independently and does not affect each other;
The unified access portal provided by GBASE can realize unified access to warehouses and clusters. For applications, the access of applications to multiple physical clusters is transparent, and the access is a unified cluster, but the original physical clusters can be planned into multiple virtual clusters according to the business system;
With permission, GBASE can access each other's virtual clusters.
Efficient Compression The efficient and transparent compression technology can automatically select the optimal compression algorithm according to the data type and data distribution rules, reduce the storage space occupied by the data as much as possible, reduce the IO consumption of the query, and improve the query performance. You can set instance-level, table-level, and column-level compression options to flexibly balance the relationship between performance and compression ratio, and the compression and decompression process is transparent to users.
Compared with traditional row-memory databases, efficient transparent compression technology can bring about an order of magnitude of performance improvement. The compression ratio can reach 2 20 times or even better, which is much higher than that of row storage;
Save 50%-95% of storage space and greatly reduce energy consumption for data processing;
Built-in compression algorithms of different levels;
In the compressed state, the requirements for IO are greatly reduced, and the data loading and query performance are significantly improved. Smart indexing
Intelligent index is a kind of coarse-grained index, every 65536 rows of data generates a data packet, each data packet automatically establishes an intelligent index when the data is stored, contains filtering information and statistical information, and can get statistical values without unpacking when data is queried, which can further reduce the IO, and the optimization effect of complex queries is obvious. Intelligent indexes in tables are automatically created and do not need to be manually created and maintained.
The intelligent index itself occupies very little space, and the scalability is very good, and the storage space is almost no expansion after the intelligent index is established.
The speed of intelligent index creation is fast, and the speed of intelligent index creation of later packets will not be affected by the previous packets.
Compared to traditional database indexing technologies, intelligent indexes are built on data packets (coarse-grained indexes) and each field is automatically indexed, while traditional indexes are built on every row of data (fine-grained indexes), so accessing intelligent indexes requires less iO (tens of thousandths of a thousandth) than accessing traditional indexes. At the same time, intelligent indexes take up about 1% of the data space, compared to 20% to 50% of the data in traditional database indexes.