From: Time series big data: a survey on data stream frameworks, analysis and algorithms
Framework | Latency | Throughput | Scalability | Fault tolerance |
---|---|---|---|---|
Apache Hadoop | High | High | High | Replication in the HDFS |
Apache Spark | Low | High | High | Resilient Distributed Dataset |
Apache Flink | Low | High | High | Incremental checkpointing (uses markers) |
Apache Storm | Low | High | High | Record-level acknowledgements |
Apache Heron | Low | High | High | High fault tolerance |
Apache Samza | Low | High | High | Host-affinity, and incremental checkpointing |
Amazon Kinesis | Low | High | High | High fault tolerance |