Feature | Description | Comment | Status | Component | Priority | Arrival(Expected) |
---|
Kafka Source(NRT) | Ingest streaming data in batch way | In design phase, did not have a conclusion of how to implement. | | | | |
Kafka Source(Real-time OLAP) | Ingest streaming data in stream/micro-batch way | In design phase, did not have a conclusion of how to implement. | | | | |
JDBC Source(Original version) | Ingest data via JDBC contract | In design phase, did not have a conclusion of how to implement. | | | | |
JDBC Source(Datasource SDK) | Ingest data via JDBC contract | In design phase, did not have a conclusion of how to implement. | | | | |
MapReduce Build Engine | Build pre-calculated cuboid data by Hadoop Mapreduce | This feature maybe useless | | | |
|
Spark Build Engine | Build pre-calculated cuboid data by Apache Spark | New implementation provided | | | | |
Flink Build Engine | Build pre-calculated cuboid data by Apache Flink | Support in Kylin 3.1 | | | |
|
HBase Storage | Use HBase to store pre-calculated cuboid data. | Discussion in mailing list | | | |
|
Parquet Storage | Use Parquet to store pre-calculated cuboid data. | Discussion in mailing list | | | | |
Distributed Query Engine / Sparder | Use calcite&catayst(a.k.a. Spark SQL) to parse/analyse/excute a SQL query. | New implementation provided | | | | |
Measure - Bitmap | Precise count distinct. | N/A | | | | |
Measure - HLL | Non-precise count distinct but low cost. | N/A | | | | |
Measure - TopN | TopN Measure | N/A | | | | |
Measure - Percentile | Percentile | N/A | | | | |
Query Cache | Cache query result in query's memory or external cache service. | N/A | | | | |
HBase Metastore | Use HBase as metastore. | I guess it will be removed in the GA version. (xxyu) | | | | |
RDBMS Metastore | Use RDBMS as metastore. | Should as the first choice of metastore. | | | | |
Cardinality Computation | Calculate cardinality of fact table and dimension table. | Planning | | | | |
Storage Cleanup | Remove useless data from storage or metastore. | New implementation | | | | |
CSV Source | Build cube from user-side csv file. | New implementation | | | | |
SQL Standard | to be updated | In testing. | | | | |
Global Dictionary(Hive) | Use hive and MR to build global dictionary | New global dictionary will replace this feature. | | | |
|
Global Dictionary(AppendTireDictionary) | Tire dictionary | New global dictionary will replace this feature. | | |
|
|
Global Dictionary(Spark Bucket Dictionary) | Use apache spark to build global dictionary | New implementation | | | | |
Cube Planner | to be updated | In design phase, did not have a conclusion of how to implement. | | | | |
System Cube and Dashboard | to be updated | Not well tested, planning | | | | |
Read write Seperatation | The query engine and build engine use different Hadoop cluster. | New implementation provided | | | | |
Pushdown Engine | to be updated | New pushdown engine will only support SparkSQL. | | | | |
Shrunken Dictionary | to be updated | This feature maybe useless | | | |
|
UHC dictionary | to be updated | This feature maybe useless | | | |
|
Deploy on AWS EMR | Support deploy Kylin on EMR5.x, EMR 6.x . Support Glue. | Planning | | | | |
All-in-one container | Provided a quick-start container for learning purpose. | How to learn Kylin in Docker | | | | |
Hadoop3 support | Going to support Hadoop3 + Hive2 in 2020-Q4. Not sure when to suppoort Hive 3. - AWS EMR 6.X
- CDH 6.X (latest 6.3.2)
| Planning | | | | |
Hive3 support |
| N/A | | | | |
Spark3 support | Support use Spark3 for build and query. | N/A | | | | |
Hybrid Model / Flexible cuboid build | Add dimension or remove dimension without purge whole cube data. | Planning | | | | |
Multi-level partition segment | Looke like Hive's multi-level partition design. | N/A | | | | |
Use catalyst to replace calcite | Make query analysis quicker and lighter. | N/A | | | | |