To improve query speed, Delta Lake on Databricks supports the ability to optimize the layout of data stored in cloud storage. Delta Lake on Databricks supports two layout algorithms: bin-packing and Z-Ordering.
This article describes how to run the optimization commands, how the two layout algorithms work, and how to clean up stale table snapshots.
- The FAQ explains why optimization is not automatic and includes recommendations for how often to run optimize commands.
- For notebooks that demonstrate the benefits of optimization, see Optimization examples.
- For reference information on Delta Lake on Databricks SQL optimization commands, see Optimize (Delta Lake on Databricks).
Delta Lake on Databricks can improve the speed of read queries from a table by coalescing small files into larger ones. You trigger compaction by running the
If you have a large amount of data and only want to optimize a subset of it, you can specify an optional partition predicate using
OPTIMIZE events WHERE date >= '2017-01-01'
- Bin-packing optimization is idempotent, meaning that if it is run twice on the same dataset, the second run has no effect.
- Bin-packing aims to produce evenly-balanced data files with respect to their size on disk, but not necessarily number of tuples per file. However, the two measures are most often correlated.
Readers of Delta tables use snapshot isolation, which means that they are not interrupted when
OPTIMIZE removes unnecessary files from the transaction log.
OPTIMIZE makes no data related changes to the table, so a read before and after an
OPTIMIZE has the same results. Performing
OPTIMIZE on a table that is a streaming source does not affect any current or future streams that treat this table as a source.
OPTIMIZE returns the file statistics (min, max, total, and so on) for the files removed and the files added by the operation. Optimize stats also contains the Z-Ordering statistics, the number of batches, and partitions optimized.
Available in Databricks Runtime 6.0 and above.
You can also compact small files automatically using Auto Optimize.
Data skipping information is collected automatically when you write data into a Delta table. Delta Lake on Databricks takes advantage of this information (minimum and maximum values) at query time to provide faster queries. You do not need to configure data skipping; the feature is activated whenever applicable. However, its effectiveness depends on the layout of your data. For best results, apply Z-Ordering.
For an example of the benefits of Delta Lake on Databricks data skipping and Z-Ordering, see the notebooks in Optimization examples. By default Delta Lake on Databricks collects statistics on the first 32 columns defined in your table schema. You can change this value using the table property
dataSkippingNumIndexedCols. Adding more columns to collect statistics would add additional overhead as you write files.
Collecting statistics on long strings is an expensive operation. To avoid collecting statistics on long strings, you can either configure the table property
dataSkippingNumIndexedCols to avoid columns containing long strings or move columns containing long strings to a column greater than
ALTER TABLE CHANGE COLUMN. For the purposes of collecting statistics, each field within a nested column is considered as an individual column.
You can read more on this article in the blog post: Processing Petabytes of Data in Seconds with Databricks Delta.
Z-Ordering is a technique to colocate related information in the same set of files. This co-locality is automatically used by Delta Lake on Databricks data-skipping algorithms to dramatically reduce the amount of data that needs to be read. To Z-Order data, you specify the columns to order on in the
ZORDER BY clause:
OPTIMIZE events WHERE date >= current_timestamp() - INTERVAL 1 day ZORDER BY (eventType)
If you expect a column to be commonly used in query predicates and if that column has high cardinality (that is, a large number of distinct values), then use
You can specify multiple columns for
ZORDER BY as a comma-separated list. However, the effectiveness of the locality drops with each additional column. Z-Ordering on columns that do not have statistics collected on them would be ineffective and a waste of resources as data skipping requires column-local stats such as min, max, and count. You can configure statistics collection on certain columns by re-ordering columns in the schema or increasing the number of columns to collect statistics on. See the section Data skipping for more details.
Z-Ordering is not idempotent but aims to be an incremental operation. The time it takes for Z-Ordering is not guaranteed to reduce over multiple runs. However, if no new data was added to a partition that was just Z-Ordered, another Z-Ordering of that partition will not have any effect.
Z-Ordering aims to produce evenly-balanced data files with respect to the number of tuples, but not necessarily data size on disk. The two measures are most often correlated, but there can be situations when that is not the case, leading to skew in optimize task times.
For example, if you
ZORDER BYdate and your most recent records are all much wider (for example longer arrays or string values) than the ones in the past, it is expected that the
OPTIMIZEjob’s task durations will be skewed, as well as the resulting file sizes. This is, however, only a problem for the
OPTIMIZEcommand itself; it should not have any negative impact on subsequent queries.
For an example of the benefits of optimization, see the following notebooks:
Delta Engine offers a few additional mechanisms to improve query performance.
At the beginning of each query Delta tables auto-update to the latest version of the table. This process can be observed in notebooks when the command status reports:
Updating the Delta table's state. However, when running historical analysis on a table, you may not necessarily need up-to-the-last-minute data, especially for tables where streaming data is being ingested frequently. In these cases, queries can be run on stale snapshots of your Delta table. This can lower latency in getting results from queries.
You can configure how stale your table data is by setting the Spark session configuration
spark.databricks.delta.stalenessLimit with a time string value, for example,
1d for 1 hour, 15 minutes, and 1 day respectively. This configuration is session specific, therefore won’t affect other users accessing this table from other notebooks, jobs, or BI tools. In addition, this setting doesn’t prevent your table from updating; it only prevents a query from having to wait for the table to update. The update still occurs in the background, and will share resources fairly across the cluster. If the staleness limit is exceeded, then the query will block on the table state update.
Delta Lake writes checkpoints as an aggregate state of a Delta table every 10 commits. These checkpoints serve as the starting point to compute the latest state of the table. Without checkpoints, Delta Lake would have to read a large collection of JSON files (“delta” files) representing commits to the transaction log to compute the state of a table. In addition, the column-level statistics Delta Lake uses to perform data skipping are stored in the checkpoint.
Delta Lake checkpoints are different than Structured Streaming checkpoints.
In Databricks Runtime 7.2 and below, column-level statistics are stored in Delta Lake checkpoints as a JSON column. In Databricks Runtime 7.3 and above, column-level statistics as a struct. The struct format makes Delta Lake reads much faster, because:
- Delta Lake doesn’t perform expensive JSON parsing to obtain column-level statistics.
- Parquet column pruning capabilities significantly reduce the I/O required to read the statistics for a column.
The struct format enables a collection of optimizations that reduce the overhead of Delta Lake read operations from seconds to tens of milliseconds, which significantly reduces the latency for short queries.
You manage how statistics are written in checkpoints using the table properties
If both table properties are
false, Delta Lake cannot perform data skipping.
In Databricks Runtime 7.3 and above:
- Batch writes write statistics in both JSON and struct format.
- Streaming writes write statistics in only JSON format (to minimize the impact of checkpoints on write latency). To also write the struct format, see Enable enhanced checkpoints for Structured Streaming queries.
- In both cases,
delta.checkpoint.writeStatsAsStructis undefined by default.
- Readers use the struct column when available and otherwise fall back to using the JSON column.
In Databricks Runtime 7.2 and below, readers only use the JSON column. Therefore, if
false, such readers cannot perform data skipping.
Enhanced checkpoints do not break compatibility with open source Delta Lake readers. However, setting
false may have implications on proprietary Delta Lake readers. Contact your vendors to learn more about performance implications.
Since writing statistics in a checkpoint has a cost (usually < a minute even for large tables), there is a tradeoff between the time taken to write a checkpoint and compatibility with Databricks Runtime 7.2 and below. If you are able to upgrade all of your workloads to Databricks Runtime 7.3 or above you can reduce the cost of writing a checkpoint by disabling the legacy JSON statistics. This tradeoff is summarized in the following table.
If data skipping is not useful in your application, you can set both properties to false, and no statistics are collected or written. We do not recommend this configuration.
If your Structured Streaming workloads don’t have low latency requirements (sub-minute latencies), you can enable enhanced checkpoints by running the following SQL command:
ALTER TABLE [<table-name>|delta.`<path-to-table>`] SET TBLPROPERTIES ('delta.checkpoint.writeStatsAsStruct' = 'true')
If you do not use Databricks Runtime 7.2 or below to query your data, you can also improve the checkpoint write latency by setting the following table properties:
ALTER TABLE [<table-name>|delta.`<path-to-table>`] SET TBLPROPERTIES ( 'delta.checkpoint.writeStatsAsStruct' = 'true', 'delta.checkpoint.writeStatsAsJson' = 'false' )
Writers in Databricks Runtime 7.2 and below write checkpoints without the stats struct, which prevents optimizations for Databricks Runtime 7.3 readers.
To block clusters running Databricks Runtime 7.2 and below from writing to a Delta table, you can upgrade the Delta table using the
from delta.tables import DeltaTable delta = DeltaTable.forPath(spark, "path_to_table") # or DeltaTable.forName delta.upgradeTableProtocol(1, 3)
import io.delta.tables.DeltaTable val delta = DeltaTable.forPath(spark, "path_to_table") // or DeltaTable.forName delta.upgradeTableProtocol(1, 3)
upgradeTableProtocol method prevents clusters running Databricks Runtime 7.2 and below from writing to your table and this change is irreversible.
We recommend upgrading your tables only after you are committed to the new format. You can try out these optimizations by creating a shallow CLONE of your tables using Databricks Runtime 7.3.
Once you upgrade the table writer version, writers must obey your settings for
The following table summarizes how to take advantage of enhanced checkpoints in various versions of Databricks Runtime, table protocol versions, and writer types.
|Without Protocol Upgrade||With Protocol Upgrade|
|Databricks Runtime 7.2 and below writer||Databricks Runtime 7.3 and above batch writer||Databricks Runtime 7.3 and above streaming writer||Databricks Runtime 7.2 and below writer||Databricks Runtime 7.3 and above batch writer||Databricks Runtime 7.3 and above streaming writer|
|Databricks Runtime 7.2 and below reader performance||No improvement||No improvement||No improvement||Cannot use writer||No improvement||No improvement|
|Databricks Runtime 7.3 and above reader performance||No improvement||Improved by default||Opt-in by table property (1)||Cannot use writer||Improved by default||Opt-in by table property (1)|
(1) Set the table property
'delta.checkpoint.writeStatsAsStruct' = 'true'
OPTIMIZE operation starts up many Spark jobs in order to optimize the file sizing via compaction (and optionally perform Z-Ordering). Since much of what
OPTIMIZE does is compact small files, you must first accumulate many small files before this operation has an effect. Therefore, the
OPTIMIZE operation is not run automatically.
OPTIMIZE, especially with
ZORDER, is an expensive operation in time and resources. If Databricks ran
OPTIMIZE automatically or waited to write out data in batches, it would remove the ability to run low-latency Delta Lake streams (where a Delta table is the source). Many customers have Delta tables that are never optimized because they only stream data from these tables, obviating the query benefits that
OPTIMIZE would provide.
Lastly, Delta Lake automatically collects statistics about the files that are written to the table (whether through an
OPTIMIZE operation or not). This means that reads from Delta tables leverage this information whether or not the table or a partition has had the
OPTIMIZE operation run on it.
How often should I run
When you choose how often to run
OPTIMIZE, there is a trade-off between performance and cost. You should run
OPTIMIZE more often if you want better end-user query performance (necessarily at a higher cost because of resource usage). You should run it less often if you want to optimize cost.
We recommend you start by running
OPTIMIZE on a daily basis (preferably at night when spot prices are low). Then modify your job from there.
What’s the best instance type to run
OPTIMIZE (bin-packing and Z-Ordering) on?
Both operations are CPU intensive operations doing large amounts of Parquet decoding and encoding.
For these workloads we recommend the c5d series.