Configuring Cloud Cache

Dremio Cloud Cache parameters are defined in dremio.conf.

Supported Data Sources

Dremio supports cloud-caching for Parquet files on the following data sources:

• Amazon S3
• Azure Storage (ADLS Gen 2) - v2 only
• Google Cloud Storage
• HDFS
• Hive on S3, Azure Storage, GCS, and HDFS

Enable Caching

Enable cloud-caching for supported data sources either when adding a new data source to your deployment or later by editing the data source.

To enable cloud-caching:

1. On the Datasets page, select a supported data source in the Data Lakes list.

2. In the top-right corner of the page, click the gear icon.

3. In the Edit Source dialog, follow these steps:

   1. Select Advanced Options, and then select Enable asynchronous access when possible.

   2. Under Cache Options, select Enable local caching when possible.

   3. (Optional) In the Max percent of total available cache space to use when possible field, specify the maximum percentage of cache space that can be used by a data source on any single executor node. The default is 100% of the total disk space available on the mount point provided for caching.

   4. Click Save.

Configuring the Cache Path and Directory

The following can be specified:

• executor.cache.path.db (optional) Database directory path - This is the path for the database directory to use for storing cached data. If you do not specify a database path, paths.local directory path is used.
• executor.cache.path.fs (optional) File system path - This is the file system cache directory. If you do not specify a file system path, the paths.local path is used.

Cache is lost if the directory location is changed or the executor is restarted.

note

Dremio recommends a cloud cache directory size of at least 100 GB. See Additition Storage for more information.

Example

In the following dremio.conf example, a database path and four (4) file system paths are specified.

If these paths were not specified, the cache manager uses the local path /mnt/resource/dremio/data. Dremio uses 70 percent of the total available disk space for the specified database and file system mount paths.

Example Code Cache configuration

paths: {
    # the local path for dremio to store data.
    local: "/mnt/resource/dremio/data"

    # the distributed path Dremio data including job results, downloads, uploads, etc
    #dist: "file://"${paths.local}"/pdfs"
}

services: {
    coordinator.enabled: false,
    coordinator.master.enabled: false,
    executor.enabled: true,
    executor.cache.path.db : "/mnt/cachemanagerdisk/db",
    executor.cache.path.fs : [ "/mnt/cachemanagerdisk/dir1","/mnt/cachemanagerdisk/dir2","/mnt/cachemanagerdisk/dir3","/mnt/cachemanagerdisk/dir4"]
}

zookeeper: "lak-azure-perf:"${services.coordinator.master.embedded-zookeeper.port}

Setting Up Caching for Reflection Data

You can improve the performance of queries that use reflections by enabling caching of reflection data.

If you are using a cloud storage provider, such as AWS, Google Cloud Platform, or Microsoft Azure, as a distributed store, caching is enabled by default. If you want to disable it, add the support key reflection.cloud.cache.enabled and set it to false. See Support Keys to learn how to add a support key.

If you are using HDFS as a distributed store, uncomment dist.caching.enabled in the debug section of dremio.conf, and set it to true. Then, restart the cluster.

Best Practices

The following describes best practices for cloud-caching:

• Use SSD/NVMe disks for best performance.
• Provide disks with sufficient space to benefit from caching. Note that the cache will evict unused data when required.
• To add capacity to the cache, add additional disks to the executor.cache.path.fs property in the dremio.conf file. Note that removing a disk is not supported.
• When removing local caching on executor nodes, you need to:
  1. Remove the Cloud Cache options on the data source.
  2. Delete the Cache Manager database and file system folders.