ClickHouse/docs/en/operations/storing-data.md

slug	sidebar_position	sidebar_label	title
/en/operations/storing-data	68	External Disks for Storing Data	External Disks for Storing Data

Data, processed in ClickHouse, is usually stored in the local file system — on the same machine with the ClickHouse server. That requires large-capacity disks, which can be expensive enough. To avoid that you can store the data remotely — on Amazon S3 disks or in the Hadoop Distributed File System (HDFS).

To work with data stored on Amazon S3 disks use S3 table engine, and to work with data in the Hadoop Distributed File System — HDFS table engine.

To load data from a web server with static files use a disk with type web.

Configuring HDFS

MergeTree and Log family table engines can store data to HDFS using a disk with type HDFS.

Configuration markup:

<clickhouse>
    <storage_configuration>
        <disks>
            <hdfs>
                <type>hdfs</type>
                <endpoint>hdfs://hdfs1:9000/clickhouse/</endpoint>
            </hdfs>
        </disks>
        <policies>
            <hdfs>
                <volumes>
                    <main>
                        <disk>hdfs</disk>
                    </main>
                </volumes>
            </hdfs>
        </policies>
    </storage_configuration>

    <merge_tree>
        <min_bytes_for_wide_part>0</min_bytes_for_wide_part>
    </merge_tree>
</clickhouse>

Required parameters:

• endpoint — HDFS endpoint URL in path format. Endpoint URL should contain a root path to store data.

Optional parameters:

• min_bytes_for_seek — The minimal number of bytes to use seek operation instead of sequential read. Default value: 1 Mb.

Using Virtual File System for Data Encryption

You can encrypt the data stored on S3, or HDFS external disks, or on a local disk. To turn on the encryption mode, in the configuration file you must define a disk with the type encrypted and choose a disk on which the data will be saved. An encrypted disk ciphers all written files on the fly, and when you read files from an encrypted disk it deciphers them automatically. So you can work with an encrypted disk like with a normal one.

Example of disk configuration:

<disks>
  <disk1>
    <type>local</type>
    <path>/path1/</path>
  </disk1>
  <disk2>
    <type>encrypted</type>
    <disk>disk1</disk>
    <path>path2/</path>
    <key>_16_ascii_chars_</key>
  </disk2>
</disks>

For example, when ClickHouse writes data from some table to a file store/all_1_1_0/data.bin to disk1, then in fact this file will be written to the physical disk along the path /path1/store/all_1_1_0/data.bin.

When writing the same file to disk2, it will actually be written to the physical disk at the path /path1/path2/store/all_1_1_0/data.bin in encrypted mode.

Required parameters:

• type — encrypted. Otherwise the encrypted disk is not created.
• disk — Type of disk for data storage.
• key — The key for encryption and decryption. Type: Uint64. You can use key_hex parameter to encode the key in hexadecimal form. You can specify multiple keys using the id attribute (see example above).

Optional parameters:

• path — Path to the location on the disk where the data will be saved. If not specified, the data will be saved in the root directory.
• current_key_id — The key used for encryption. All the specified keys can be used for decryption, and you can always switch to another key while maintaining access to previously encrypted data.
• algorithm — Algorithm for encryption. Possible values: AES_128_CTR, AES_192_CTR or AES_256_CTR. Default value: AES_128_CTR. The key length depends on the algorithm: AES_128_CTR — 16 bytes, AES_192_CTR — 24 bytes, AES_256_CTR — 32 bytes.

Example of disk configuration:

<clickhouse>
    <storage_configuration>
        <disks>
            <disk_s3>
                <type>s3</type>
                <endpoint>...
            </disk_s3>
            <disk_s3_encrypted>
                <type>encrypted</type>
                <disk>disk_s3</disk>
                <algorithm>AES_128_CTR</algorithm>
                <key_hex id="0">00112233445566778899aabbccddeeff</key_hex>
                <key_hex id="1">ffeeddccbbaa99887766554433221100</key_hex>
                <current_key_id>1</current_key_id>
            </disk_s3_encrypted>
        </disks>
    </storage_configuration>
</clickhouse>

Using local cache

It is possible to configure local cache over disks in storage configuration starting from version 22.3. For versions 22.3 - 22.7 cache is supported only for s3 disk type. For versions >= 22.8 cache is supported for any disk type: S3, Azure, Local, Encrypted, etc. For versions >= 23.5 cache is supported only for remote disk types: S3, Azure, HDFS. Cache uses LRU cache policy.

Example of configuration for versions later or equal to 22.8:

<clickhouse>
    <storage_configuration>
        <disks>
            <s3>
                <type>s3</type>
                <endpoint>...</endpoint>
                ... s3 configuration ...
            </s3>
            <cache>
                <type>cache</type>
                <disk>s3</disk>
                <path>/s3_cache/</path>
                <max_size>10Gi</max_size>
            </cache>
        </disks>
        <policies>
            <s3-cache>
                <volumes>
                    <main>
                        <disk>cache</disk>
                    </main>
                </volumes>
            </s3-cache>
        <policies>
    </storage_configuration>

Example of configuration for versions earlier than 22.8:

<clickhouse>
    <storage_configuration>
        <disks>
            <s3>
                <type>s3</type>
                <endpoint>...</endpoint>
                ... s3 configuration ...
                <data_cache_enabled>1</data_cache_enabled>
                <data_cache_max_size>10737418240</data_cache_max_size>
            </s3>
        </disks>
        <policies>
            <s3-cache>
                <volumes>
                    <main>
                        <disk>s3</disk>
                    </main>
                </volumes>
            </s3-cache>
        <policies>
    </storage_configuration>

File Cache disk configuration settings:

These settings should be defined in the disk configuration section.

• path - path to the directory with cache. Default: None, this setting is obligatory.

• max_size - maximum size of the cache in bytes or in readable format, e.g. ki, Mi, Gi, etc, example 10Gi (such format works starting from 22.10 version). When the limit is reached, cache files are evicted according to the cache eviction policy. Default: None, this setting is obligatory.

• cache_on_write_operations - allow to turn on write-through cache (caching data on any write operations: INSERT queries, background merges). Default: false. The write-through cache can be disabled per query using setting enable_filesystem_cache_on_write_operations (data is cached only if both cache config settings and corresponding query setting are enabled).

• enable_filesystem_query_cache_limit - allow to limit the size of cache which is downloaded within each query (depends on user setting max_query_cache_size). Default: false.

• enable_cache_hits_threshold - number which defines how many times some data needs to be read before it will be cached. Default: false. This threshold can be defined by cache_hits_threshold. Default: 0, e.g. the data is cached at the first attempt to read it.

• enable_bypass_cache_with_threshold - allows to skip cache completely in case the requested read range exceeds the threshold. Default: false. This threshold can be defined by bypass_cache_threashold. Default: 268435456 (256Mi).

• max_file_segment_size - a maximum size of a single cache file in bytes or in readable format (ki, Mi, Gi, etc, example 10Gi). Default: 8388608 (8Mi).

• max_elements - a limit for a number of cache files. Default: 10000000.

• load_metadata_threads - number of threads being used to load cache metadata on starting time. Default: 16.

File Cache query/profile settings:

Some of these settings will disable cache features per query/profile that are enabled by default or in disk configuration settings. For example, you can enable cache in disk configuration and disable it per query/profile setting enable_filesystem_cache to false. Also setting cache_on_write_operations to true in disk configuration means that "write-though" cache is enabled. But if you need to disable this general setting per specific queries then setting enable_filesystem_cache_on_write_operations to false means that write operations cache will be disabled for a specific query/profile.

• enable_filesystem_cache - allows to disable cache per query even if storage policy was configured with cache disk type. Default: true.

• read_from_filesystem_cache_if_exists_otherwise_bypass_cache - allows to use cache in query only if it already exists, otherwise query data will not be written to local cache storage. Default: false.

• enable_filesystem_cache_on_write_operations - turn on write-through cache. This setting works only if setting cache_on_write_operations in cache configuration is turned on. Default: false. Cloud default value: true.

• enable_filesystem_cache_log - turn on logging to system.filesystem_cache_log table. Gives a detailed view of cache usage per query. It can be turn on for specific queries or enabled in a profile. Default: false.

• max_query_cache_size - a limit for the cache size, which can be written to local cache storage. Requires enabled enable_filesystem_query_cache_limit in cache configuration. Default: false.

• skip_download_if_exceeds_query_cache - allows to change the behaviour of setting max_query_cache_size. Default: true. If this setting is turned on and cache download limit during query was reached, no more cache will be downloaded to cache storage. If this setting is turned off and cache download limit during query was reached, cache will still be written by cost of evicting previously downloaded (within current query) data, e.g. second behaviour allows to preserve last recently used behaviour while keeping query cache limit.

Warning Cache configuration settings and cache query settings correspond to the latest ClickHouse version, for earlier versions something might not be supported.

Cache system tables:

• system.filesystem_cache - system tables which shows current state of cache.

• system.filesystem_cache_log - system table which shows detailed cache usage per query. Requires enable_filesystem_cache_log setting to be true.

Cache commands:

• SYSTEM DROP FILESYSTEM CACHE (<cache_name>) (ON CLUSTER) -- ON CLUSTER is only supported when no <cache_name> is provided

• SHOW FILESYSTEM CACHES -- show list of filesystem caches which were configured on the server. (For versions <= 22.8 the command is named SHOW CACHES)

SHOW FILESYSTEM CACHES

Result:

┌─Caches────┐
│ s3_cache  │
└───────────┘

• DESCRIBE FILESYSTEM CACHE '<cache_name>' - show cache configuration and some general statistics for a specific cache. Cache name can be taken from SHOW FILESYSTEM CACHES command. (For versions <= 22.8 the command is named DESCRIBE CACHE)

DESCRIBE FILESYSTEM CACHE 's3_cache'

┌────max_size─┬─max_elements─┬─max_file_segment_size─┬─boundary_alignment─┬─cache_on_write_operations─┬─cache_hits_threshold─┬─current_size─┬─current_elements─┬─path───────┬─background_download_threads─┬─enable_bypass_cache_with_threshold─┐
│ 10000000000 │      1048576 │             104857600 │            4194304 │                         1 │                    0 │         3276 │               54 │ /s3_cache/ │                           2 │                                  0 │
└─────────────┴──────────────┴───────────────────────┴────────────────────┴───────────────────────────┴──────────────────────┴──────────────┴──────────────────┴────────────┴─────────────────────────────┴────────────────────────────────────┘

Cache current metrics:

• FilesystemCacheSize

• FilesystemCacheElements

Cache asynchronous metrics:

• FilesystemCacheBytes

• FilesystemCacheFiles

Cache profile events:

• CachedReadBufferReadFromSourceBytes, CachedReadBufferReadFromCacheBytes,

• CachedReadBufferReadFromSourceMicroseconds, CachedReadBufferReadFromCacheMicroseconds

• CachedReadBufferCacheWriteBytes, CachedReadBufferCacheWriteMicroseconds

• CachedWriteBufferCacheWriteBytes, CachedWriteBufferCacheWriteMicroseconds

Using in-memory cache (userspace page cache)

The File Cache described above stores cached data in local files. Alternatively, object-store-based disks can be configured to use "Userspace Page Cache", which is RAM-only. Userspace page cache is recommended only if file cache can't be used for some reason, e.g. if the machine doesn't have a local disk at all. Note that file cache effectively uses RAM for caching too, since the OS caches contents of local files.

To enable userspace page cache for disks that don't use file cache, use setting use_page_cache_for_disks_without_file_cache.

By default, on Linux, the userspace page cache will use all available memory, similar to the OS page cache. In tools like top and ps, the clickhouse server process will typically show resident set size near 100% of the machine's RAM - this is normal, and most of this memory is actually reclaimable by the OS on memory pressure (MADV_FREE). This behavior can be disabled with server setting page_cache_use_madv_free = 0, making the userspace page cache just use a fixed amount of memory page_cache_size with no special interaction with the OS. On Mac OS, page_cache_use_madv_free is always disabled as it doesn't have lazy MADV_FREE.

Unfortunately, page_cache_use_madv_free makes it difficult to tell if the server is close to running out of memory, since the RSS metric becomes useless. Async metric UnreclaimableRSS shows the amount of physical memory used by the server, excluding the memory reclaimable by the OS: select value from system.asynchronous_metrics where metric = 'UnreclaimableRSS'. Use it for monitoring instead of RSS. This metric is only available if page_cache_use_madv_free is enabled.

Storing Data on Web Server

There is a tool clickhouse-static-files-uploader, which prepares a data directory for a given table (SELECT data_paths FROM system.tables WHERE name = 'table_name'). For each table you need, you get a directory of files. These files can be uploaded to, for example, a web server with static files. After this preparation, you can load this table into any ClickHouse server via DiskWeb.

This is a read-only disk. Its data is only read and never modified. A new table is loaded to this disk via ATTACH TABLE query (see example below). Local disk is not actually used, each SELECT query will result in a http request to fetch required data. All modification of the table data will result in an exception, i.e. the following types of queries are not allowed: CREATE TABLE, ALTER TABLE, RENAME TABLE, DETACH TABLE and TRUNCATE TABLE.

Web server storage is supported only for the MergeTree and Log engine families. To access the data stored on a web disk, use the storage_policy setting when executing the query. For example, ATTACH TABLE table_web UUID '{}' (id Int32) ENGINE = MergeTree() ORDER BY id SETTINGS storage_policy = 'web'.

A ready test case. You need to add this configuration to config:

<clickhouse>
    <storage_configuration>
        <disks>
            <web>
                <type>web</type>
                <endpoint>https://clickhouse-datasets.s3.yandex.net/disk-with-static-files-tests/test-hits/</endpoint>
            </web>
        </disks>
        <policies>
            <web>
                <volumes>
                    <main>
                        <disk>web</disk>
                    </main>
                </volumes>
            </web>
        </policies>
    </storage_configuration>
</clickhouse>

And then execute this query:

ATTACH TABLE test_hits UUID '1ae36516-d62d-4218-9ae3-6516d62da218'
(
    WatchID UInt64,
    JavaEnable UInt8,
    Title String,
    GoodEvent Int16,
    EventTime DateTime,
    EventDate Date,
    CounterID UInt32,
    ClientIP UInt32,
    ClientIP6 FixedString(16),
    RegionID UInt32,
    UserID UInt64,
    CounterClass Int8,
    OS UInt8,
    UserAgent UInt8,
    URL String,
    Referer String,
    URLDomain String,
    RefererDomain String,
    Refresh UInt8,
    IsRobot UInt8,
    RefererCategories Array(UInt16),
    URLCategories Array(UInt16),
    URLRegions Array(UInt32),
    RefererRegions Array(UInt32),
    ResolutionWidth UInt16,
    ResolutionHeight UInt16,
    ResolutionDepth UInt8,
    FlashMajor UInt8,
    FlashMinor UInt8,
    FlashMinor2 String,
    NetMajor UInt8,
    NetMinor UInt8,
    UserAgentMajor UInt16,
    UserAgentMinor FixedString(2),
    CookieEnable UInt8,
    JavascriptEnable UInt8,
    IsMobile UInt8,
    MobilePhone UInt8,
    MobilePhoneModel String,
    Params String,
    IPNetworkID UInt32,
    TraficSourceID Int8,
    SearchEngineID UInt16,
    SearchPhrase String,
    AdvEngineID UInt8,
    IsArtifical UInt8,
    WindowClientWidth UInt16,
    WindowClientHeight UInt16,
    ClientTimeZone Int16,
    ClientEventTime DateTime,
    SilverlightVersion1 UInt8,
    SilverlightVersion2 UInt8,
    SilverlightVersion3 UInt32,
    SilverlightVersion4 UInt16,
    PageCharset String,
    CodeVersion UInt32,
    IsLink UInt8,
    IsDownload UInt8,
    IsNotBounce UInt8,
    FUniqID UInt64,
    HID UInt32,
    IsOldCounter UInt8,
    IsEvent UInt8,
    IsParameter UInt8,
    DontCountHits UInt8,
    WithHash UInt8,
    HitColor FixedString(1),
    UTCEventTime DateTime,
    Age UInt8,
    Sex UInt8,
    Income UInt8,
    Interests UInt16,
    Robotness UInt8,
    GeneralInterests Array(UInt16),
    RemoteIP UInt32,
    RemoteIP6 FixedString(16),
    WindowName Int32,
    OpenerName Int32,
    HistoryLength Int16,
    BrowserLanguage FixedString(2),
    BrowserCountry FixedString(2),
    SocialNetwork String,
    SocialAction String,
    HTTPError UInt16,
    SendTiming Int32,
    DNSTiming Int32,
    ConnectTiming Int32,
    ResponseStartTiming Int32,
    ResponseEndTiming Int32,
    FetchTiming Int32,
    RedirectTiming Int32,
    DOMInteractiveTiming Int32,
    DOMContentLoadedTiming Int32,
    DOMCompleteTiming Int32,
    LoadEventStartTiming Int32,
    LoadEventEndTiming Int32,
    NSToDOMContentLoadedTiming Int32,
    FirstPaintTiming Int32,
    RedirectCount Int8,
    SocialSourceNetworkID UInt8,
    SocialSourcePage String,
    ParamPrice Int64,
    ParamOrderID String,
    ParamCurrency FixedString(3),
    ParamCurrencyID UInt16,
    GoalsReached Array(UInt32),
    OpenstatServiceName String,
    OpenstatCampaignID String,
    OpenstatAdID String,
    OpenstatSourceID String,
    UTMSource String,
    UTMMedium String,
    UTMCampaign String,
    UTMContent String,
    UTMTerm String,
    FromTag String,
    HasGCLID UInt8,
    RefererHash UInt64,
    URLHash UInt64,
    CLID UInt32,
    YCLID UInt64,
    ShareService String,
    ShareURL String,
    ShareTitle String,
    ParsedParams Nested(
        Key1 String,
        Key2 String,
        Key3 String,
        Key4 String,
        Key5 String,
        ValueDouble Float64),
    IslandID FixedString(16),
    RequestNum UInt32,
    RequestTry UInt8
)
ENGINE = MergeTree()
PARTITION BY toYYYYMM(EventDate)
ORDER BY (CounterID, EventDate, intHash32(UserID))
SAMPLE BY intHash32(UserID)
SETTINGS storage_policy='web';

Required parameters:

• type — web. Otherwise the disk is not created.
• endpoint — The endpoint URL in path format. Endpoint URL must contain a root path to store data, where they were uploaded.

Optional parameters:

• min_bytes_for_seek — The minimal number of bytes to use seek operation instead of sequential read. Default value: 1 Mb.
• remote_fs_read_backoff_threashold — The maximum wait time when trying to read data for remote disk. Default value: 10000 seconds.
• remote_fs_read_backoff_max_tries — The maximum number of attempts to read with backoff. Default value: 5.

If a query fails with an exception DB:Exception Unreachable URL, then you can try to adjust the settings: http_connection_timeout, http_receive_timeout, keep_alive_timeout.

To get files for upload run: clickhouse static-files-disk-uploader --metadata-path <path> --output-dir <dir> (--metadata-path can be found in query SELECT data_paths FROM system.tables WHERE name = 'table_name').

When loading files by endpoint, they must be loaded into <endpoint>/store/ path, but config must contain only endpoint.

If URL is not reachable on disk load when the server is starting up tables, then all errors are caught. If in this case there were errors, tables can be reloaded (become visible) via DETACH TABLE table_name -> ATTACH TABLE table_name. If metadata was successfully loaded at server startup, then tables are available straight away.

Use http_max_single_read_retries setting to limit the maximum number of retries during a single HTTP read.

Zero-copy Replication (not ready for production)

Zero-copy replication is possible, but not recommended, with S3 and HDFS disks. Zero-copy replication means that if the data is stored remotely on several machines and needs to be synchronized, then only the metadata is replicated (paths to the data parts), but not the data itself.

:::note Zero-copy replication is not ready for production Zero-copy replication is disabled by default in ClickHouse version 22.8 and higher. This feature is not recommended for production use. :::