Help Center> Distributed Database Middleware> Glossary> Glossary> Glossary

A-E

    • A-E
      DDM

      See Distributed Database Middleware

      DDM instance

      A DDM instance is the minimum resource unit of DDM, which runs on an ECS. DDM instances currently support four types of specifications, including 4 vCPUs | 8 GB, 8 vCPUs | 16 GB, 16 vCPUs | 32 GB, 32 vCPUs | 64 GB, 48 vCPUs | 96 GB, and 64 vCPUs | 128 GB. You can apply for suitable DDM instances according to service requirements.

      Distributed Database Middleware

      Distributed Database Middleware (DDM) is designed to solve distributed expansion issues of databases. It breaks the capacity and performance bottlenecks of traditional databases to achieve high concurrent access to mass data.

      DDM uses the Huawei Relational Database Service (RDS) as the storage engine and provides full-lifecycle O&M management capabilities, such as automatic deployment, sharding, flexible scaling, and high availability.

    F-J

      • F
        Full table scan

        If a sharding column is not specified in SQL for a sharded table, SQL will be performed on all shards. Such a process is called a full table scan.

        You are advised to optimize SQL statements and specify a sharding column to avoid full table scans.

      • G
        Global sequence

        In a distributed database, data is stored in each sharded table, and the primary key constraints in a relational database are unavailable.

        For this reason, the global sequence number is used for DDM to ensure that data retains the uniqueness after sharded tables are aggregated.

        Global table

        In a distributed database, data is stored in each sharded table, and the primary key constraints in a relational database are unavailable. For this reason, the global sequence number is used for DDM to ensure that data retains the uniqueness after sharded tables are aggregated.

      • H
        Horizontal sharding

        Data tables are sharded into multiple sharded tables based on sharding rules. After the sharding, data is stored in the sharded tables based on sharding rules.

        Horizontal sharding raises the capacity limit of a table in a single-node database and improves access performance.

      P-T

        • R
          Read/Write isolation

          DDM redirects read and write requests of application services to different RDS DB instances.

          When the system has a heavy read load, read replicas are added to RDS DB instances. Data is read from read replicas, improving the concurrent database performance.

          Application codes do not need to be changed because the read/write isolation of DDM is open to applications.

        • S
          Seamless scale-out

          The database storage space is expanded by increasing RDS DB instances without service interruption.

          Capacity expansion breaks the storage capacity bottleneck and improves the database performance by enhancing the concurrent computing capability.

          The DDM console supports capacity expansion. During capacity expansion, applications are not reconstructed and the progress is visualized.

          Shard and sharded table

          A DDM instance creates multiple physical databases in an RDS DB instance. Each physical database is a shard. A table that is created in a shard is a sharded table, which is neither a non-sharded table nor a global table.

          Data is distributed to shards based on sharding rules. The data supports distributed concurrent computing.

        U-Z

          • U-Z
            Unsharded table

            Tables that have a small data volume and do not need to be queried by connecting to sharded tables can be created to store data only in the default shards of DDM.

            These tables are called non-sharded tables. A non-sharded table can be created only by connecting to the schema through the MySQL client or application programs.