ETL Writeback Acceleration
Download
Modo Foco
Tamanho da Fonte
TDSQL Boundless provides an accelerated ETL write-back feature. This feature uses a columnar read-only analysis instance to accelerate the query performance of the SELECT clause in
INSERT ... SELECT ... statements. The query results are then written back to the target table in the read/write instance over the internal network, significantly improving ETL efficiency in complex query scenarios.Technical Principles
As shown in the figure above, you can route the
INSERT ... SELECT ... statement to the columnar read-only analysis instance for execution. The SELECT clause is accelerated by the columnar read-only analysis instance, and the query results are directly written back to the read/write instance. This improves overall execution efficiency.Use Cases
We recommend using this feature in the following scenarios: complex query conditions, long SQL execution times, but a small result set. In such scenarios, the columnar read-only analysis instance can significantly accelerate the execution efficiency of SELECT queries, thereby improving overall ETL performance.
This feature does not guarantee performance gains in all scenarios. Performance may degrade in the following cases:
When the SELECT query in the
INSERT ... SELECT ... statement is simple, reading data from the columnar read-only analysis instance and writing it back to the read/write instance incurs additional network overhead. Compared to reading data directly from the read/write instance, the advantage is not significant.When the SELECT query result set in the
INSERT ... SELECT ... statement is large, the main performance bottleneck lies in the process of transmitting the result set over the network and writing it to the read/write instance. This feature cannot optimize performance in this scenario.Attention:
The columnar read-only analysis instance uses asynchronous replication. When sync latency exists, a time difference occurs between the query results and the read/write instance. This feature is only suitable for scenarios that are not sensitive to data latency.
Ajuda e Suporte
Esta página foi útil?
Você também pode entrar em contato com a Equipe de vendas ou Enviar um tíquete em caso de ajuda.
comentários