Comparisons to Other Products¶
Hologres V.S. Real-time/Time-series Database¶
Druid, Clickhouse,¶
HBase, Cassandra¶
Kudu¶
- Must be used with Impala
- Point-lookup is slower than HBase and Cassandra, and analytics is slower than Druid and ClickHouse
Hologres V.S. Data Lake¶
Data Lake setups like Apache Iceberg, Apache Hudi, Delta Lake are mostly libraries plus a cloud distributed file system. They are poor man, open-source version of commercial data warehouses.
Given they are libraries maintaining data on top of commodity storage, Data Lakes have to be used with Spark or Presto.
Fits better with mini-batch processing engine like Spark, but not Flink No native streaming support Data latency depends on how frequently files are commited No point-lookup support They are libaraies not services, thus their capabilities are limited and operation cost is very high. E.g. No compaction to compact small files as their number grows No TTL support