GaussDB(DWS) can seamlessly interconnect with HDFS storage, allowing you to query data on the Hadoop platform through foreign tables. GaussDB(DWS) can perform big data analysis on ORC, TXT, CSV, and Parquet files.

      The shared-nothing MPP architecture can fully utilize resources, and its performance improves linearly as the cluster is scaled out, supporting up to petabytes of data. Connecting GaussDB(DWS) to Hadoop has the following advantages:

      - Compatibility with standard SQL

        GaussDB(DWS) is compatible with SQL92, 99, and 2003 standards and stored procedures, which reduces the cost of application development and migration.

      - Comprehensive transaction support

        You can add, delete, and modify ORC data in HDFS. Atomicity, consistency, isolation, and durability (ACID) properties are guaranteed by MVCC and two-phase commit (2PC) distributed transaction management. The Smart Insert hybrid row and column storage engine enables intelligent data allocation, storage, and high-speed update.

      - High-performance interactive query

        Using the MPP asynchronous pipeline distributed execution framework, vectorized executor, predicate pushdown, intelligent filtering and scanning, and low level virtual machine (LLVM) technologies, the database achieves industry-leading kernel engine performance and interactive query and analysis performance.

      The amount of data in all sorts of industries has been increasing sharply. A single data center does not have enough storage and computing capabilities to handle it all. During data mining and correlation analysis, a query may involve data from multiple data centers.

      GaussDB(DWS) can directly access other databases, including Oracle, MySQL, and Hadoop, facilitating collaborative analysis across databases.

Figure 5-1 Cross-DC collaborative analysis

      GaussDB(DWS) can work with Huawei Cloud OBS, a cloud native storage service. Using the Express cluster acceleration technology, GaussDB(DWS) can access data in OBS at a high speed, archive historical data, and separate cold and hot data.

      In the Safe City project, AI models are used for image recognition and vehicle search based on the multi-dimensional data comparison and analysis capabilities provided by OBS and GaussDB(DWS), as shown in the following figure.

Figure 5-2 Fast image recognition based on AI and cloud native technologies