• State Backends
    • Available State Backends
      • The MemoryStateBackend
      • The FsStateBackend
      • The RocksDBStateBackend
    • Configuring a State Backend
      • Setting the Per-job State Backend
      • Setting Default State Backend
        • RocksDB State Backend Config Options

    State Backends

    Programs written in the Data Stream API often hold state in various forms:

    • Windows gather elements or aggregates until they are triggered
    • Transformation functions may use the key/value state interface to store values
    • Transformation functions may implement the CheckpointedFunction interface to make their local variables fault tolerant

    See also state section in the streaming API guide.

    When checkpointing is activated, such state is persisted upon checkpoints to guard against data loss and recover consistently.How the state is represented internally, and how and where it is persisted upon checkpoints depends on thechosen State Backend.

    • Available State Backends
      • The MemoryStateBackend
      • The FsStateBackend
      • The RocksDBStateBackend
    • Configuring a State Backend
      • Setting the Per-job State Backend
      • Setting Default State Backend

    Available State Backends

    Out of the box, Flink bundles these state backends:

    • MemoryStateBackend
    • FsStateBackend
    • RocksDBStateBackend

    If nothing else is configured, the system will use the MemoryStateBackend.

    The MemoryStateBackend

    The MemoryStateBackend holds data internally as objects on the Java heap. Key/value state and window operators hold hash tablesthat store the values, triggers, etc.

    Upon checkpoints, this state backend will snapshot the state and send it as part of the checkpoint acknowledgement messages to theJobManager (master), which stores it on its heap as well.

    The MemoryStateBackend can be configured to use asynchronous snapshots. While we strongly encourage the use of asynchronous snapshots to avoid blocking pipelines, please note that this is currently enabled by default. To disable this feature, users can instantiate a MemoryStateBackend with the corresponding boolean flag in the constructor set to false(this should only used for debug), e.g.:

    1. new MemoryStateBackend(MAX_MEM_STATE_SIZE, false);

    Limitations of the MemoryStateBackend:

    • The size of each individual state is by default limited to 5 MB. This value can be increased in the constructor of the MemoryStateBackend.
    • Irrespective of the configured maximal state size, the state cannot be larger than the akka frame size (see Configuration).
    • The aggregate state must fit into the JobManager memory.

    The MemoryStateBackend is encouraged for:

    • Local development and debugging
    • Jobs that do hold little state, such as jobs that consist only of record-at-a-time functions (Map, FlatMap, Filter, …). The Kafka Consumer requires very little state.

    The FsStateBackend

    The FsStateBackend is configured with a file system URL (type, address, path), such as “hdfs://namenode:40010/flink/checkpoints” or “file:///data/flink/checkpoints”.

    The FsStateBackend holds in-flight data in the TaskManager’s memory. Upon checkpointing, it writes state snapshots into files in the configured file system and directory. Minimal metadata is stored in the JobManager’s memory (or, in high-availability mode, in the metadata checkpoint).

    The FsStateBackend uses asynchronous snapshots by default to avoid blocking the processing pipeline while writing state checkpoints. To disable this feature, users can instantiate a FsStateBackend with the corresponding boolean flag in the constructor set to false, e.g.:

    1. new FsStateBackend(path, false);

    The FsStateBackend is encouraged for:

    • Jobs with large state, long windows, large key/value states.
    • All high-availability setups.

    The RocksDBStateBackend

    The RocksDBStateBackend is configured with a file system URL (type, address, path), such as “hdfs://namenode:40010/flink/checkpoints” or “file:///data/flink/checkpoints”.

    The RocksDBStateBackend holds in-flight data in a RocksDB databasethat is (per default) stored in the TaskManager data directories. Upon checkpointing, the wholeRocksDB database will be checkpointed into the configured file system and directory. Minimalmetadata is stored in the JobManager’s memory (or, in high-availability mode, in the metadata checkpoint).

    The RocksDBStateBackend always performs asynchronous snapshots.

    Limitations of the RocksDBStateBackend:

    • As RocksDB’s JNI bridge API is based on byte[], the maximum supported size per key and per value is 2^31 bytes each. IMPORTANT: states that use merge operations in RocksDB (e.g. ListState) can silently accumulate value sizes > 2^31 bytes and will then fail on their next retrieval. This is currently a limitation of RocksDB JNI.

    The RocksDBStateBackend is encouraged for:

    • Jobs with very large state, long windows, large key/value states.
    • All high-availability setups.

    Note that the amount of state that you can keep is only limited by the amount of disk space available.This allows keeping very large state, compared to the FsStateBackend that keeps state in memory.This also means, however, that the maximum throughput that can be achieved will be lower withthis state backend. All reads/writes from/to this backend have to go through de-/serialization to retrieve/store the state objects, which is also more expensive than always working with theon-heap representation as the heap-based backends are doing.

    RocksDBStateBackend is currently the only backend that offers incremental checkpoints (see here).

    Certain RocksDB native metrics are available but disabled by default, you can find full documentation here

    Configuring a State Backend

    The default state backend, if you specify nothing, is the jobmanager. If you wish to establish a different default for all jobs on your cluster, you can do so by defining a new default state backend in flink-conf.yaml. The default state backend can be overridden on a per-job basis, as shown below.

    Setting the Per-job State Backend

    The per-job state backend is set on the StreamExecutionEnvironment of the job, as shown in the example below:

    1. StreamExecutionEnvironment env = StreamExecutionEnvironment.getExecutionEnvironment();
    2. env.setStateBackend(new FsStateBackend("hdfs://namenode:40010/flink/checkpoints"));
    1. val env = StreamExecutionEnvironment.getExecutionEnvironment()
    2. env.setStateBackend(new FsStateBackend("hdfs://namenode:40010/flink/checkpoints"))

    If you want to use the RocksDBStateBackend, then you have to add the following dependency to your Flink project.

    1. <dependency>
    2. <groupId>org.apache.flink</groupId>
    3. <artifactId>flink-statebackend-rocksdb_2.11</artifactId>
    4. <version>1.9.0</version>
    5. </dependency>

    Setting Default State Backend

    A default state backend can be configured in the flink-conf.yaml, using the configuration key state.backend.

    Possible values for the config entry are jobmanager (MemoryStateBackend), filesystem (FsStateBackend), rocksdb (RocksDBStateBackend), or the fully qualified classname of the class that implements the state backend factory StateBackendFactory,such as org.apache.flink.contrib.streaming.state.RocksDBStateBackendFactory for RocksDBStateBackend.

    The state.checkpoints.dir option defines the directory to which all backends write checkpoint data and meta data files.You can find more details about the checkpoint directory structure here.

    A sample section in the configuration file could look as follows:

    1. # The backend that will be used to store operator state checkpoints
    2. state.backend: filesystem
    3. # Directory for storing checkpoints
    4. state.checkpoints.dir: hdfs://namenode:40010/flink/checkpoints

    RocksDB State Backend Config Options

    KeyDefaultDescription
    ##### state.backend.rocksdb.checkpoint.transfer.thread.num1The number of threads (per stateful operator) used to transfer (download and upload) files in RocksDBStateBackend.
    ##### state.backend.rocksdb.localdir(none)The local directory (on the TaskManager) where RocksDB puts its files.
    ##### state.backend.rocksdb.options-factory"org.apache.flink.contrib.streaming.state.DefaultConfigurableOptionsFactory"The options factory class for RocksDB to create DBOptions and ColumnFamilyOptions. The default options factory is org.apache.flink.contrib.streaming.state.DefaultConfigurableOptionsFactory, and it would read the configured options which provided in 'RocksDBConfigurableOptions'.
    ##### state.backend.rocksdb.predefined-options"DEFAULT"The predefined settings for RocksDB DBOptions and ColumnFamilyOptions by Flink community. Current supported candidate predefined-options are DEFAULT, SPINNING_DISK_OPTIMIZED, SPINNING_DISK_OPTIMIZED_HIGH_MEM or FLASH_SSD_OPTIMIZED. Note that user customized options and options from the OptionsFactory are applied on top of these predefined ones.
    ##### state.backend.rocksdb.timer-service.factory"HEAP"This determines the factory for timer service state implementation. Options are either HEAP (heap-based, default) or ROCKSDB for an implementation based on RocksDB .
    ##### state.backend.rocksdb.ttl.compaction.filter.enabledfalseThis determines if compaction filter to cleanup state with TTL is enabled for backend.Note: User can still decide in state TTL configuration in state descriptor whether the filter is active for particular state or not.