Basic Configurations
This page covers the basic configurations you may use to write/read Hudi tables. This page only features a subset of the most frequently used configurations. For a full list of all configs, please visit the All Configurations page.
- Spark Datasource Configs: These configs control the Hudi Spark Datasource, providing ability to define keys/partitioning, pick out the write operation, specify how to merge records or choosing query type to read.
- Flink Sql Configs: These configs control the Hudi Flink SQL source/sink connectors, providing ability to define record keys, pick out the write operation, specify how to merge records, enable/disable asynchronous compaction or choosing query type to read.
- Write Client Configs: Internally, the Hudi datasource uses a RDD based HoodieWriteClient API to actually perform writes to storage. These configs provide deep control over lower level aspects like file sizing, compression, parallelism, compaction, write schema, cleaning etc. Although Hudi provides sane defaults, from time-time these configs may need to be tweaked to optimize for specific workloads.
- Metrics Configs: These set of configs are used to enable monitoring and reporting of key Hudi stats and metrics.
- Record Payload Config: This is the lowest level of customization offered by Hudi. Record payloads define how to produce new values to upsert based on incoming new record and stored old record. Hudi provides default implementations such as OverwriteWithLatestAvroPayload which simply update table with the latest/last-written record. This can be overridden to a custom class extending HoodieRecordPayload class, on both datasource and WriteClient levels.
Spark Datasource Configs
These configs control the Hudi Spark Datasource, providing ability to define keys/partitioning, pick out the write operation, specify how to merge records or choosing query type to read.
Read Options
Options useful for reading tables via read.format.option(...)
Config Class: org.apache.hudi.DataSourceOptions.scala
hoodie.datasource.query.type
Whether data needs to be read, in incremental mode (new data since an instantTime) (or) Read Optimized mode (obtain latest view, based on base files) (or) Snapshot mode (obtain latest view, by merging base and (if any) log files)
Default Value: snapshot (Optional)
Config Param: QUERY_TYPE
Write Options
You can pass down any of the WriteClient level configs directly using options() or option(k,v) methods.
inputDF.write()
.format("org.apache.hudi")
.options(clientOpts) // any of the Hudi client opts can be passed in as well
.option(DataSourceWriteOptions.RECORDKEY_FIELD_OPT_KEY(), "_row_key")
.option(DataSourceWriteOptions.PARTITIONPATH_FIELD_OPT_KEY(), "partition")
.option(DataSourceWriteOptions.PRECOMBINE_FIELD_OPT_KEY(), "timestamp")
.option(HoodieWriteConfig.TABLE_NAME, tableName)
.mode(SaveMode.Append)
.save(basePath);
Options useful for writing tables via write.format.option(...)
Config Class: org.apache.hudi.DataSourceOptions.scala
hoodie.datasource.write.operation
Whether to do upsert, insert or bulkinsert for the write operation. Use bulkinsert to load new data into a table, and there after use upsert/insert. bulk insert uses a disk based write path to scale to load large inputs without need to cache it.
Default Value: upsert (Optional)
Config Param: OPERATION
hoodie.datasource.write.table.type
The table type for the underlying data, for this write. This can’t change between writes.
Default Value: COPY_ON_WRITE (Optional)
Config Param: TABLE_TYPE
hoodie.datasource.write.table.name
Table name for the datasource write. Also used to register the table into meta stores.
Default Value: N/A (Required)
Config Param: TABLE_NAME
hoodie.datasource.write.recordkey.field
Record key field. Value to be used as the
recordKeycomponent ofHoodieKey. Actual value will be obtained by invoking .toString() on the field value. Nested fields can be specified using the dot notation eg:a.b.c
Default Value: uuid (Optional)
Config Param: RECORDKEY_FIELD
hoodie.datasource.write.partitionpath.field
Partition path field. Value to be used at the partitionPath component of HoodieKey. Actual value ontained by invoking .toString()
Default Value: N/A (Required)
Config Param: PARTITIONPATH_FIELD
hoodie.datasource.write.keygenerator.class
Key generator class, that implements
org.apache.hudi.keygen.KeyGenerator
Default Value: org.apache.hudi.keygen.SimpleKeyGenerator (Optional)
Config Param: KEYGENERATOR_CLASS_NAME
hoodie.datasource.write.precombine.field
Field used in preCombining before actual write. When two records have the same key value, we will pick the one with the largest value for the precombine field, determined by Object.compareTo(..)
Default Value: ts (Optional)
Config Param: PRECOMBINE_FIELD
hoodie.datasource.write.payload.class
Payload class used. Override this, if you like to roll your own merge logic, when upserting/inserting. This will render any value set for PRECOMBINE_FIELD_OPT_VAL in-effective
Default Value: org.apache.hudi.common.model.OverwriteWithLatestAvroPayload (Optional)
Config Param: PAYLOAD_CLASS_NAME
hoodie.datasource.write.partitionpath.urlencode
Should we url encode the partition path value, before creating the folder structure.
Default Value: false (Optional)
Config Param: URL_ENCODE_PARTITIONING
hoodie.datasource.hive_sync.enable
When set to true, register/sync the table to Apache Hive metastore
Default Value: false (Optional)
Config Param: HIVE_SYNC_ENABLED
hoodie.datasource.hive_sync.mode
Mode to choose for Hive ops. Valid values are hms, jdbc and hiveql.
Default Value: N/A (Required)
Config Param: HIVE_SYNC_MODE
hoodie.datasource.write.hive_style_partitioning
Flag to indicate whether to use Hive style partitioning. If set true, the names of partition folders follow <partition_column_name>=<partition_value> format. By default false (the names of partition folders are only partition values)
Default Value: false (Optional)
Config Param: HIVE_STYLE_PARTITIONING
hoodie.datasource.hive_sync.partition_fields
Field in the table to use for determining hive partition columns.
Default Value: (Optional)
Config Param: HIVE_PARTITION_FIELDS
hoodie.datasource.hive_sync.partition_extractor_class
Class which implements PartitionValueExtractor to extract the partition values, default 'SlashEncodedDayPartitionValueExtractor'.
Default Value: org.apache.hudi.hive.SlashEncodedDayPartitionValueExtractor (Optional)
Config Param: HIVE_PARTITION_EXTRACTOR_CLASS
Flink Sql Configs
These configs control the Hudi Flink SQL source/sink connectors, providing ability to define record keys, pick out the write operation, specify how to merge records, enable/disable asynchronous compaction or choosing query type to read.
Flink Options
path
Base path for the target hoodie table. The path would be created if it does not exist, otherwise a Hoodie table expects to be initialized successfully
Default Value: N/A (Required)
Config Param: PATH
hoodie.table.name
Table name to register to Hive metastore
Default Value: N/A (Required)
Config Param: TABLE_NAME
table.type
Type of table to write. COPY_ON_WRITE (or) MERGE_ON_READ
Default Value: COPY_ON_WRITE (Optional)
Config Param: TABLE_TYPE
write.operation
The write operation, that this write should do
Default Value: upsert (Optional)
Config Param: OPERATION
write.tasks
Parallelism of tasks that do actual write, default is 4
Default Value: 4 (Optional)
Config Param: WRITE_TASKS
write.bucket_assign.tasks
Parallelism of tasks that do bucket assign, default is the parallelism of the execution environment
Default Value: N/A (Required)
Config Param: BUCKET_ASSIGN_TASKS
write.precombine
Flag to indicate whether to drop duplicates before insert/upsert. By default these cases will accept duplicates, to gain extra performance:
- insert operation;
- upsert for MOR table, the MOR table deduplicate on reading
Default Value: false (Optional)
Config Param: PRE_COMBINE
read.tasks
Parallelism of tasks that do actual read, default is 4
Default Value: 4 (Optional)
Config Param: READ_TASKS
read.start-commit
Start commit instant for reading, the commit time format should be 'yyyyMMddHHmmss', by default reading from the latest instant for streaming read
Default Value: N/A (Required)
Config Param: READ_START_COMMIT
read.streaming.enabled
Whether to read as streaming source, default false
Default Value: false (Optional)
Config Param: READ_AS_STREAMING
compaction.tasks
Parallelism of tasks that do actual compaction, default is 4
Default Value: 4 (Optional)
Config Param: COMPACTION_TASKS
hoodie.datasource.write.hive_style_partitioning
Whether to use Hive style partitioning. If set true, the names of partition folders follow <partition_column_name>=<partition_value> format. By default false (the names of partition folders are only partition values)
Default Value: false (Optional)
Config Param: HIVE_STYLE_PARTITIONING
hive_sync.enable
Asynchronously sync Hive meta to HMS, default false
Default Value: false (Optional)
Config Param: HIVE_SYNC_ENABLED
hive_sync.mode
Mode to choose for Hive ops. Valid values are hms, jdbc and hiveql, default 'jdbc'
Default Value: jdbc (Optional)
Config Param: HIVE_SYNC_MODE
hive_sync.table
Table name for hive sync, default 'unknown'
Default Value: unknown (Optional)
Config Param: HIVE_SYNC_TABLE
hive_sync.db
Database name for hive sync, default 'default'
Default Value: default (Optional)
Config Param: HIVE_SYNC_DB
hive_sync.partition_extractor_class
Tool to extract the partition value from HDFS path, default 'SlashEncodedDayPartitionValueExtractor'
Default Value: org.apache.hudi.hive.SlashEncodedDayPartitionValueExtractor (Optional)
Config Param: HIVE_SYNC_PARTITION_EXTRACTOR_CLASS_NAME
hive_sync.metastore.uris
Metastore uris for hive sync, default ''
Default Value: (Optional)
Config Param: HIVE_SYNC_METASTORE_URIS
Write Client Configs
Internally, the Hudi datasource uses a RDD based HoodieWriteClient API to actually perform writes to storage. These configs provide deep control over lower level aspects like file sizing, compression, parallelism, compaction, write schema, cleaning etc. Although Hudi provides sane defaults, from time-time these configs may need to be tweaked to optimize for specific workloads.
Storage Configs
Configurations that control aspects around writing, sizing, reading base and log files.
Config Class: org.apache.hudi.config.HoodieStorageConfig
write.parquet.block.size
Parquet RowGroup size. It's recommended to make this large enough that scan costs can be amortized by packing enough column values into a single row group.
Default Value: 120 (Optional)
Config Param: WRITE_PARQUET_BLOCK_SIZE
write.parquet.max.file.size
Target size for parquet files produced by Hudi write phases. For DFS, this needs to be aligned with the underlying filesystem block size for optimal performance.
Default Value: 120 (Optional)
Config Param: WRITE_PARQUET_MAX_FILE_SIZE
Metadata Configs
Configurations used by the Hudi Metadata Table. This table maintains the metadata about a given Hudi table (e.g file listings) to avoid overhead of accessing cloud storage, during queries.
Config Class: org.apache.hudi.common.config.HoodieMetadataConfig
hoodie.metadata.enable
Enable the internal metadata table which serves table metadata like level file listings
Default Value: true (Optional)
Config Param: ENABLE
Since Version: 0.7.0
Write Configurations
Configurations that control write behavior on Hudi tables. These can be directly passed down from even higher level frameworks (e.g Spark datasources, Flink sink) and utilities (e.g DeltaStreamer).
Config Class: org.apache.hudi.config.HoodieWriteConfig
hoodie.combine.before.upsert
When upserted records share same key, controls whether they should be first combined (i.e de-duplicated) before writing to storage. This should be turned off only if you are absolutely certain that there are no duplicates incoming, otherwise it can lead to duplicate keys and violate the uniqueness guarantees.
Default Value: true (Optional)
Config Param: COMBINE_BEFORE_UPSERT
hoodie.write.markers.type
Marker type to use. Two modes are supported: - DIRECT: individual marker file corresponding to each data file is directly created by the writer. - TIMELINE_SERVER_BASED: marker operations are all handled at the timeline service which serves as a proxy. New marker entries are batch processed and stored in a limited number of underlying files for efficiency. If HDFS is used or timeline server is disabled, DIRECT markers are used as fallback even if this is configured. For Spark structured streaming, this configuration does not take effect, i.e., DIRECT markers are always used for Spark structured streaming.
Default Value: TIMELINE_SERVER_BASED (Optional)
Config Param: MARKERS_TYPE
Since Version: 0.9.0
hoodie.insert.shuffle.parallelism
Parallelism for inserting records into the table. Inserts can shuffle data before writing to tune file sizes and optimize the storage layout.
Default Value: 200 (Optional)
Config Param: INSERT_PARALLELISM_VALUE
hoodie.rollback.parallelism
Parallelism for rollback of commits. Rollbacks perform delete of files or logging delete blocks to file groups on storage in parallel.
Default Value: 100 (Optional)
Config Param: ROLLBACK_PARALLELISM_VALUE
hoodie.combine.before.delete
During delete operations, controls whether we should combine deletes (and potentially also upserts) before writing to storage.
Default Value: true (Optional)
Config Param: COMBINE_BEFORE_DELETE
hoodie.combine.before.insert
When inserted records share same key, controls whether they should be first combined (i.e de-duplicated) before writing to storage. When set to true the precombine field value is used to reduce all records that share the same key.
Default Value: false (Optional)
Config Param: COMBINE_BEFORE_INSERT
hoodie.bulkinsert.shuffle.parallelism
For large initial imports using bulk_insert operation, controls the parallelism to use for sort modes or custom partitioning done before writing records to the table.
Default Value: 200 (Optional)
Config Param: BULKINSERT_PARALLELISM_VALUE
hoodie.delete.shuffle.parallelism
Parallelism used for “delete” operation. Delete operations also perform shuffles, similar to upsert operation.
Default Value: 200 (Optional)
Config Param: DELETE_PARALLELISM_VALUE
hoodie.bulkinsert.sort.mode
Sorting modes to use for sorting records for bulk insert. This is used when user hoodie.bulkinsert.user.defined.partitioner.class is not configured. Available values are - GLOBAL_SORT: this ensures best file sizes, with lowest memory overhead at cost of sorting. PARTITION_SORT: Strikes a balance by only sorting within a partition, still keeping the memory overhead of writing lowest and best effort file sizing. NONE: No sorting. Fastest and matches
spark.write.parquet()in terms of number of files, overheads
Default Value: GLOBAL_SORT (Optional)
Config Param: BULK_INSERT_SORT_MODE
hoodie.embed.timeline.server
When true, spins up an instance of the timeline server (meta server that serves cached file listings, statistics),running on each writer's driver process, accepting requests during the write from executors.
Default Value: true (Optional)
Config Param: EMBEDDED_TIMELINE_SERVER_ENABLE
hoodie.upsert.shuffle.parallelism
Parallelism to use for upsert operation on the table. Upserts can shuffle data to perform index lookups, file sizing, bin packing records optimally into file groups.
Default Value: 200 (Optional)
Config Param: UPSERT_PARALLELISM_VALUE
hoodie.rollback.using.markers
Enables a more efficient mechanism for rollbacks based on the marker files generated during the writes. Turned on by default.
Default Value: true (Optional)
Config Param: ROLLBACK_USING_MARKERS_ENABLE
hoodie.finalize.write.parallelism
Parallelism for the write finalization internal operation, which involves removing any partially written files from lake storage, before committing the write. Reduce this value, if the high number of tasks incur delays for smaller tables or low latency writes.
Default Value: 200 (Optional)
Config Param: FINALIZE_WRITE_PARALLELISM_VALUE
Compaction Configs
Configurations that control compaction (merging of log files onto new base files) as well as cleaning (reclamation of older/unused file groups/slices).
Config Class: org.apache.hudi.config.HoodieCompactionConfig
hoodie.cleaner.policy
Cleaning policy to be used. The cleaner service deletes older file slices files to re-claim space. By default, cleaner spares the file slices written by the last N commits, determined by hoodie.cleaner.commits.retained Long running query plans may often refer to older file slices and will break if those are cleaned, before the query has had a chance to run. So, it is good to make sure that the data is retained for more than the maximum query execution time
Default Value: KEEP_LATEST_COMMITS (Optional)
Config Param: CLEANER_POLICY
hoodie.copyonwrite.record.size.estimate
The average record size. If not explicitly specified, hudi will compute the record size estimate dynamically based on commit metadata. This is critical in computing the insert parallelism and bin-packing inserts into small files.
Default Value: 1024 (Optional)
Config Param: COPY_ON_WRITE_RECORD_SIZE_ESTIMATE
hoodie.compact.inline.max.delta.seconds
Number of elapsed seconds after the last compaction, before scheduling a new one.
Default Value: 3600 (Optional)
Config Param: INLINE_COMPACT_TIME_DELTA_SECONDS
hoodie.cleaner.commits.retained
Number of commit to retain when cleaner is triggered with KEEP_LATEST_COMMITS cleaning policy. Make sure to configure this property properly so that the longest running query is able to succeed. This also directly translates into how much data retention the table supports for incremental queries. Default Value: 10 (Optional)
Config Param: CLEANER_COMMITS_RETAINED
hoodie.clean.async
Only applies when hoodie.clean.automatic is turned on. When turned on runs cleaner async with writing, which can speed up overall write performance.
Default Value: false (Optional)
Config Param: ASYNC_CLEAN
hoodie.clean.automatic
When enabled, the cleaner table service is invoked immediately after each commit, to delete older file slices. It's recommended to enable this, to ensure metadata and data storage growth is bounded.
Default Value: true (Optional)
Config Param: AUTO_CLEAN
hoodie.commits.archival.batch
Archiving of instants is batched in best-effort manner, to pack more instants into a single archive log. This config controls such archival batch size.
Default Value: 10 (Optional)
Config Param: COMMITS_ARCHIVAL_BATCH_SIZE
hoodie.compact.inline
When set to true, compaction service is triggered after each write. While being simpler operationally, this adds extra latency on the write path.
Default Value: false (Optional)
Config Param: INLINE_COMPACT
hoodie.parquet.small.file.limit
During upsert operation, we opportunistically expand existing small files on storage, instead of writing new files, to keep number of files to an optimum. This config sets the file size limit below which a file on storage becomes a candidate to be selected as such a
small file. By default, treat any file <= 100MB as a small file.
Default Value: 104857600 (Optional)
Config Param: PARQUET_SMALL_FILE_LIMIT
hoodie.compaction.strategy
Compaction strategy decides which file groups are picked up for compaction during each compaction run. By default, Hudi picks the log file with most accumulated unmerged data
Default Value: org.apache.hudi.table.action.compact.strategy.LogFileSizeBasedCompactionStrategy (Optional)
Config Param: COMPACTION_STRATEGY
hoodie.archive.automatic
When enabled, the archival table service is invoked immediately after each commit, to archive commits if we cross a maximum value of commits. It's recommended to enable this, to ensure number of active commits is bounded.
Default Value: true (Optional)
Config Param: AUTO_ARCHIVE
hoodie.copyonwrite.insert.auto.split
Config to control whether we control insert split sizes automatically based on average record sizes. It's recommended to keep this turned on, since hand tuning is otherwise extremely cumbersome.
Default Value: true (Optional)
Config Param: COPY_ON_WRITE_AUTO_SPLIT_INSERTS
hoodie.compact.inline.max.delta.commits
Number of delta commits after the last compaction, before scheduling of a new compaction is attempted. This is used when the compaction trigger strategy involves number of commits. For example NUM_COMMITS,NUM_AND_TIME,NUM_OR_TIME
Default Value: 5 (Optional)
Config Param: INLINE_COMPACT_NUM_DELTA_COMMITS
hoodie.keep.min.commits
Similar to hoodie.keep.max.commits, but controls the minimum number of instants to retain in the active timeline.
Default Value: 20 (Optional)
Config Param: MIN_COMMITS_TO_KEEP
hoodie.cleaner.parallelism
Parallelism for the cleaning operation. Increase this if cleaning becomes slow.
Default Value: 200 (Optional)
Config Param: CLEANER_PARALLELISM_VALUE
hoodie.record.size.estimation.threshold
We use the previous commits' metadata to calculate the estimated record size and use it to bin pack records into partitions. If the previous commit is too small to make an accurate estimation, Hudi will search commits in the reverse order, until we find a commit that has totalBytesWritten larger than (PARQUET_SMALL_FILE_LIMIT_BYTES * this_threshold)
Default Value: 1.0 (Optional)
Config Param: RECORD_SIZE_ESTIMATION_THRESHOLD
hoodie.compact.inline.trigger.strategy
Controls how compaction scheduling is triggered, by time or num delta commits or combination of both. Valid options: NUM_COMMITS,TIME_ELAPSED,NUM_AND_TIME,NUM_OR_TIME
Default Value: NUM_COMMITS (Optional)
Config Param: INLINE_COMPACT_TRIGGER_STRATEGY
hoodie.keep.max.commits
Archiving service moves older entries from timeline into an archived log after each write, to keep the metadata overhead constant, even as the table size grows.This config controls the maximum number of instants to retain in the active timeline.
Default Value: 30 (Optional)
Config Param: MAX_COMMITS_TO_KEEP
hoodie.copyonwrite.insert.split.size
Number of inserts assigned for each partition/bucket for writing. We based the default on writing out 100MB files, with at least 1kb records (100K records per file), and over provision to 500K. As long as auto-tuning of splits is turned on, this only affects the first write, where there is no history to learn record sizes from.
Default Value: 500000 (Optional)
Config Param: COPY_ON_WRITE_INSERT_SPLIT_SIZE
File System View Storage Configurations
Configurations that control how file metadata is stored by Hudi, for transaction processing and queries.
Config Class: org.apache.hudi.common.table.view.FileSystemViewStorageConfig
hoodie.filesystem.view.type
File system view provides APIs for viewing the files on the underlying lake storage, as file groups and file slices. This config controls how such a view is held. Options include MEMORY,SPILLABLE_DISK,EMBEDDED_KV_STORE,REMOTE_ONLY,REMOTE_FIRST which provide different trade offs for memory usage and API request performance.
Default Value: MEMORY (Optional)
Config Param: VIEW_TYPE
hoodie.filesystem.view.secondary.type
Specifies the secondary form of storage for file system view, if the primary (e.g timeline server) is unavailable.
Default Value: MEMORY (Optional)
Config Param: SECONDARY_VIEW_TYPE
Index Configs
Configurations that control indexing behavior, which tags incoming records as either inserts or updates to older records.
Config Class: org.apache.hudi.config.HoodieIndexConfig
hoodie.index.type
Type of index to use. Default is Bloom filter. Possible options are [BLOOM | GLOBAL_BLOOM |SIMPLE | GLOBAL_SIMPLE | INMEMORY | HBASE | BUCKET]. Bloom filters removes the dependency on a external system and is stored in the footer of the Parquet Data Files
Default Value: N/A (Required)
Config Param: INDEX_TYPE
hoodie.index.bloom.fpp
Only applies if index type is BLOOM. Error rate allowed given the number of entries. This is used to calculate how many bits should be assigned for the bloom filter and the number of hash functions. This is usually set very low (default: 0.000000001), we like to tradeoff disk space for lower false positives. If the number of entries added to bloom filter exceeds the configured value (hoodie.index.bloom.num_entries), then this fpp may not be honored.
Default Value: 0.000000001 (Optional)
Config Param: BLOOM_FILTER_FPP_VALUE
hoodie.index.bloom.num_entries
Only applies if index type is BLOOM. This is the number of entries to be stored in the bloom filter. The rationale for the default: Assume the maxParquetFileSize is 128MB and averageRecordSize is 1kb and hence we approx a total of 130K records in a file. The default (60000) is roughly half of this approximation. Warning: Setting this very low, will generate a lot of false positives and index lookup will have to scan a lot more files than it has to and setting this to a very high number will increase the size every base file linearly (roughly 4KB for every 50000 entries). This config is also used with DYNAMIC bloom filter which determines the initial size for the bloom.
Default Value: 60000 (Optional)
Config Param: BLOOM_FILTER_NUM_ENTRIES_VALUE
hoodie.bloom.index.update.partition.path
Only applies if index type is GLOBAL_BLOOM. When set to true, an update including the partition path of a record that already exists will result in inserting the incoming record into the new partition and deleting the original record in the old partition. When set to false, the original record will only be updated in the old partition
Default Value: true (Optional)
Config Param: BLOOM_INDEX_UPDATE_PARTITION_PATH_ENABLE
hoodie.bloom.index.use.caching
Only applies if index type is BLOOM. When true, the input RDD will cached to speed up index lookup by reducing IO for computing parallelism or affected partitions
Default Value: true (Optional)
Config Param: BLOOM_INDEX_USE_CACHING
hoodie.bloom.index.parallelism
Only applies if index type is BLOOM. This is the amount of parallelism for index lookup, which involves a shuffle. By default, this is auto computed based on input workload characteristics.
Default Value: 0 (Optional)
Config Param: BLOOM_INDEX_PARALLELISM
hoodie.bloom.index.prune.by.ranges
Only applies if index type is BLOOM. When true, range information from files to leveraged speed up index lookups. Particularly helpful, if the key has a monotonously increasing prefix, such as timestamp. If the record key is completely random, it is better to turn this off, since range pruning will only add extra overhead to the index lookup.
Default Value: true (Optional)
Config Param: BLOOM_INDEX_PRUNE_BY_RANGES
hoodie.bloom.index.filter.type
Filter type used. Default is BloomFilterTypeCode.DYNAMIC_V0. Available values are [BloomFilterTypeCode.SIMPLE , BloomFilterTypeCode.DYNAMIC_V0]. Dynamic bloom filters auto size themselves based on number of keys.
Default Value: DYNAMIC_V0 (Optional)
Config Param: BLOOM_FILTER_TYPE
hoodie.simple.index.parallelism
Only applies if index type is SIMPLE. This is the amount of parallelism for index lookup, which involves a Spark Shuffle
Default Value: 50 (Optional)
Config Param: SIMPLE_INDEX_PARALLELISM
hoodie.simple.index.use.caching
Only applies if index type is SIMPLE. When true, the incoming writes will cached to speed up index lookup by reducing IO for computing parallelism or affected partitions
Default Value: true (Optional)
Config Param: SIMPLE_INDEX_USE_CACHING
hoodie.global.simple.index.parallelism
Only applies if index type is GLOBAL_SIMPLE. This is the amount of parallelism for index lookup, which involves a Spark Shuffle
Default Value: 100 (Optional)
Config Param: GLOBAL_SIMPLE_INDEX_PARALLELISM
hoodie.simple.index.update.partition.path
Similar to Key: 'hoodie.bloom.index.update.partition.path' , default: true but for simple index. Since version: 0.6.0
Default Value: true (Optional)
Config Param: SIMPLE_INDEX_UPDATE_PARTITION_PATH_ENABLE
Common Configurations
The following set of configurations are common across Hudi.
Config Class: org.apache.hudi.common.config.HoodieCommonConfig
hoodie.common.spillable.diskmap.type
When handling input data that cannot be held in memory, to merge with a file on storage, a spillable diskmap is employed. By default, we use a persistent hashmap based loosely on bitcask, that offers O(1) inserts, lookups. Change this to
ROCKS_DBto prefer using rocksDB, for handling the spill.
Default Value: BITCASK (Optional)
Config Param: SPILLABLE_DISK_MAP_TYPE
Metrics Configs
These set of configs are used to enable monitoring and reporting of key Hudi stats and metrics.
Metrics Configurations for Datadog reporter
Enables reporting on Hudi metrics using the Datadog reporter type. Hudi publishes metrics on every commit, clean, rollback etc.
Config Class: org.apache.hudi.config.metrics.HoodieMetricsDatadogConfig
hoodie.metrics.on
Turn on/off metrics reporting. off by default.
Default Value: false (Optional)
Config Param: TURN_METRICS_ON
Since Version: 0.5.0
hoodie.metrics.reporter.type
Type of metrics reporter.
Default Value: GRAPHITE (Optional)
Config Param: METRICS_REPORTER_TYPE_VALUE
Since Version: 0.5.0
hoodie.metrics.reporter.class
Default Value: (Optional)
Config Param: METRICS_REPORTER_CLASS_NAME
Since Version: 0.6.0
Record Payload Config
This is the lowest level of customization offered by Hudi. Record payloads define how to produce new values to upsert based on incoming new record and stored old record. Hudi provides default implementations such as OverwriteWithLatestAvroPayload which simply update table with the latest/last-written record. This can be overridden to a custom class extending HoodieRecordPayload class, on both datasource and WriteClient levels.
Payload Configurations
Payload related configs, that can be leveraged to control merges based on specific business fields in the data.
Config Class: org.apache.hudi.config.HoodiePayloadConfig
hoodie.payload.event.time.field
Table column/field name to derive timestamp associated with the records. This can be useful for e.g, determining the freshness of the table.
Default Value: ts (Optional)
Config Param: EVENT_TIME_FIELD
hoodie.payload.ordering.field
Table column/field name to order records that have the same key, before merging and writing to storage.
Default Value: ts (Optional)
Config Param: ORDERING_FIELD