Skip to main content

Aerospike Backup (asbackup)

The asbackup utility backs up data from an Aerospike database according to a user-defined backup scope of specific namespaces, sets, or both. The scope also supports further refinement with partition or time-based filters.

Once the scope is defined, asbackup scans the database, fetching the records that match the specified criteria. This operation is similar to performing a Primary Index (PI) query, which can quickly locate and retrieve records within the defined parameters. This ensures that the backup operation captures the essential data needed for recovery and ignores non-critical system or secondary data.

As records are identified, asbackup serializes the data into a predefined format and writes it to a backup file or directory. This serialization process involves converting the in-memory representation of records into a stable format that can be safely stored on disk.

Backup contents

By default, the following data is backed up:

  • Keys
    • Key metadata: digest, TTL, generation count, and key
    • Regular bins: string, integer, boolean, and binary
    • Collection data type (CDT) bins: list and map
    • GeoJSON data type bins
    • HyperLogLog data type bins
  • Secondary index definitions
  • User-Defined Function (UDF) modules

Required permissions

The asbackup utility requires the read access role or higher.


When running in directory mode, each worker thread creates its own backup file. If the backup file is not full after completing the query, that is, less than --file-limit MB in size, the backup file is placed on a queue to be reused by another backup job.

Basic backup commands

The simplest way to run asbackup is to specify only:

  • the cluster to back up (--host)
  • the namespace to back up (--namespace). asbackup backs up a single namespace at a time.
  • the local directory for the backup files (--directory)

For example, a cluster contains a node with IP address To back up the test namespace on this cluster to the directory backup_2024_08_24, run the following command:

asbackup --host --namespace test --directory backup_2024_08_24

Data is stored in multiple files with the .asb file extension. By default, each backup file is limited to 250 MiB. When this limit is reached, asbackup creates a new file. For the exact backup file format, see the file format specification at the Backup File Format repository on GitHub.

asbackup --host --namespace test --directory ./output_dir --compress zstd

In this example, asbackup writes data from the test namespace as multiple files to the output_dir directory, compressing them using the ZSTD algorithm.

Run with the -Z or --help option to see an overview of all supported command-line options.

Estimate the backup size

asbackup can create an estimate of the average record size by creating and measuring a temporary test backup of 10,000 records from the namespace. Get this estimate with the --estimate option, skipping --directory and --output-file. To estimate the total size of the backup file or files, multiply the estimated average record size by the number of records in the namespace and add 10% for indexes and overhead.

asbackup --namespace NAME --estimate
Formula to calculate approximate disk space for backup
Estimated average record size from --estimate
× Number of records in namespace
+ 10% of estimated record size
= approximate disk space needed for backup

Per-record filters (filter-exp, modified-after, modified-before, no-ttl-only, after-digest, and partition-list) and node-list are not accounted for in the estimate, and using these options will have no effect on the estimate.


If you pass the --estimate flag, you may not pass --parallel. The two flags are mutually exclusive.

For more information about resource requirements for backup and restore, see asbackup and asrestore resource usage.

Back up to a single file

You can back up the cluster to a single file, rather than a directory:

asbackup --host HOST --namespace NAME --output-file FILENAME

Before a backup-to-file operation is run, asbackup runs an estimate on the namespace being backed up and uses it to calculate a 99.9% confidence upper bound on the total size of the backup file. The number of estimate samples taken can be controlled with --estimate-samples, with the default being 10000, just as in backup-to-directory size estimate mode.

Compression and encryption

You can compress and encrypt backup file data before it is written to the backup file with --compress and --encrypt. Enable an option by passing it to asbackup and include your chosen algorithm.


There is one available compression algorithm:

zstdZSTD compression, from the facebook libztsd repository on GitHub.

For example:

asbackup --host HOST --namespace NAME --compress zstd --compression-level 3

The compression level, set with the optional --compression-level flag, is an integer described in the zstd manual. Set the default compression level with the ZSTD_CLEVEL_DEFAULT parameter.


These are the available encryption algorithms:

aes128AES 128-bit key-digest encryption, which uses the CTR128 algorithm to encrypt data. The SHA256 hash of the encryption key is used to generate the key used by CTR128.
aes256AES 256-bit key-digest encryption, which is again the same, only using a 256-bit digest of the key for encryption and AES256 as the base encryption algorithm.

For encryption, you must provide a private key. The private encryption key may be in PEM format (with --encryption-key-file), or a base64 encoded key passed in through an environment variable (with --encryption-key-env).

For example, using an encryption key file:

asbackup --host HOST --namespace NAME --encrypt aes128 --encryption-key-file KEY.PEM

Using an environment variable:

asbackup --host HOST --namespace NAME --encrypt aes256 --encryption-key-env PRIVATE_KEY

Replace 'PRIVATE KEY' with the contents of your private key file, between the header and footer. In the following example the key starts with b3Blb and ends with eNfNpA=:


Note that when restoring compressed/encrypted backup files, the same compression/encryption flags must be provided to asrestore.

Connection options

-h HOST1:TLSNAME1:PORT1,... or --host HOST1:TLSNAME1:PORT1,... host that acts as the entry point to the cluster. Any nodes in the cluster can be specified. The remaining nodes are discovered automatically.
-p PORT or --port PORT3000Port to connect to.
-U USER or --user USER-User name with read permission. Mandatory if the server has security enabled.
-P PASSWORD or --password-Password to authenticate the given user. The first form passes the password on the command line. The second form prompts for the password.
-A or --authINTERNALSet authentication mode when user and password are defined. Modes are (INTERNAL, EXTERNAL, EXTERNAL_INSECURE, PKI) and the default is INTERNAL. This mode must be set EXTERNAL when using LDAP.
-l or --node-list ADDR1:TLSNAME1:PORT1,...localhost:3000While --host and --port automatically discover all cluster nodes, --node-list backs up a subset of cluster nodes by first calculating the subset of partitions owned by the listed nodes, and then backing up that list of partitions. This option is mutually exclusive with --partition-list and --after-digest.
--parallel N1Maximum number of scans to run in parallel. If only one partition range is given, or the entire namespace is being backed up, the range of partitions is evenly divided by this number to be processed in parallel. Otherwise, each filter cannot be parallelized individually, so you may only achieve as much parallelism as there are partition filters.
--tls-enabledisabledIndicates a TLS connection should be used.
-S or --services-alternatefalseSet this to true to connect to Aerospike node's alternate-access-address.
--prefer-racks RACKID1,...disabledA comma separated list of rack IDs to prefer when reading records for a backup. This is useful for limiting cross datacenter network traffic.

Timeout options

--socket-timeout MS10000Socket timeout in milliseconds. If this value is 0, it is set to total-timeout. If both are 0, there is no socket idle time limit.
--total-timeout MS0Total socket timeout in milliseconds. Default is 0, that is, no timeout.
--max-retries N5Maximum number of retries before aborting the current transaction.
--sleep-between-retries MS0The amount of time to sleep between retries.

TLS options

--tls-cafile=TLS_CAFILEPath to a trusted CA certificate file.
--tls-capath=TLS_CAPATHPath to a directory of trusted CA certificates.
--tls-name=TLS_NAMEThe default TLS name used to authenticate each TLS socket connection. Note: this must also match the cluster name.
--tls-protocols=TLS_PROTOCOLSSet the TLS protocol selection criteria. This format is the same as Apache's SSL Protocol. If not specified, asrestore uses TLSv1.2 if supported. Otherwise it uses -all +TLSv1.
--tls-cipher-suite=TLS_CIPHER_SUITESet the TLS cipher selection criteria. The format is the same as OpenSSL's Cipher List Format.
--tls-keyfile=TLS_KEYFILEPath to the key for mutual authentication (if Aerospike cluster supports it).
--tls-keyfile-password=TLS_KEYFILE_PASSWORDPassword to load protected TLS-keyfile. Can be one of the following:
1) Environment variable: env:VAR
2) File: file:PATH
3) String: PASSWORD
User will be prompted on command line if --tls-keyfile-password specified and no password is given.
--tls-certfile=TLS_CERTFILE <path>Path to the chain file for mutual authentication (if Aerospike Cluster supports it).
--tls-cert-blacklist <path>Path to a certificate blocklist file. The file should contain one line for each blocklisted certificate. Each line starts with the certificate serial number expressed in hex. Each entry may optionally specify the issuer name of the certificate (serial numbers are only required to be unique per issuer). Example: 867EC87482B2 /C=US/ST=CA/O=Acme/OU=Engineering/CN=TestChainCA
--tls-crl-checkEnable CRL checking for leaf certificate. An error occurs if a valid CRL files cannot be found in TLS_CAPATH.
--tls-crl-checkallEnable CRL checking for entire certificate chain. An error occurs if a valid CRL files cannot be found in TLS_CAPATH.
--tls-log-session-infoEnable logging session information for each TLS connection.

TLS_NAME is only used when connecting with a secure TLS enabled server.

The following example creates a backup with the following parameters:

  • Cluster nodes and
  • Port 3000
  • Namespace test
  • Output directory backup_2015_08_24
  • TLS enabled


asbackup --host, --namespace test --directory backup_2015_08_24 --tls-enable --tls-cafile /cluster_name.pem --tls-protocols TLSv1.2 --tls-keyfile /cluster_name.key --tls-certfile /cluster_name.pem

Output options

-d PATH or --directory PATH-Directory to store the .asb backup files in. If the directory does not exist, it will be created before use. Mandatory, unless --output-file or --estimate is given.
-o PATH or --output-file PATH-The single file to write the backup to. - means stdout. Mandatory, unless --directory or --estimate is given.


--output-file-prefix DESIRED-PREFIX
Must be used with the --directory option. A desired prefix for all output files.
-e or --estimate-Specified in lieu of --directory or --output-file, estimates the average size of a single record in the backup file. Useful for estimating the expected size of a backup before actually starting it. Multiply the returned value by the number of records in the namespace and add 10% for overhead. This option is mutually exclusive to --remove-artifacts and --continue.
--estimate-samples N10000Sets the number of record samples to take in a backup estimate. This also sets the number of estimate samples taken for the estimate run before backup-to-file.
-F LIMIT or --file-limit LIMIT250 MiBFile size limit (in MiB) for --directory. If a .asb backup file crosses this size threshold, asbackup will switch to a new file.
-r or --remove-files-Clear directory or remove output file. By default, asbackup refuses to write to a non-empty directory or to overwrite an existing backup file. This option clears the given --directory or removes an existing --output-file. Mutually exclusive to --continue.
--remove-artifacts-Clear directory or remove output file, like --remove-files, without running a backup. This option is mutually exclusive to --continue and --estimate.
-C or --compact-Do not base-64 encode BLOB values. For better readability of backup files, asbackup base-64 encodes BLOB values by default. This option disables the encoding step, which saves space in the backup file. However, be prepared to encounter odd-looking binary data in your backup files.
-N BANDWIDTH or --nice BANDWIDTH-Throttles asbackup's write operations to the backup file(s) to not exceed the given bandwidth in MiB/s. Effectively also throttles the scan on the server side as asbackup refuses to accept more data than it can write.
-y ENCRYPTION-ALG or --encrypt ENCRYPTION-ALGnoneThe encryption algorithm to be used on backup files as they are written. The options available are aes128 and aes256. This option must be accompanied by either --encryption-key-file or --encryption-key-env. Refer to compression and encryption
-z COMPRESSION-ALG or --compress COMPRESSION-ALGnoneThe compression algorithm to be used on backup files as they are written. The options available are zstd. Refer to compression and encryption
--compression-level N3The zstd compression level to be used. Refer to the zstd manual for more information.

Incremental backup

The argument YYYY-MMM-DD_HH:MM:SS is the time stamp variable:

  • -a or --modified-after YYYY-MMM-DD_HH:MM:SS backs up keys time-stamped after the argument.
  • -b or --modified-before YYYY-MMM-DD_HH:MM:SS backs up keys time-stamped before the argument.

You may also back up partitions to create incremental backups. Refer to partition lists.

Namespace data selection options

-n NAMESPACE or --namespace NAMESPACE-Namespace to backup. Mandatory.
-s SETS or --set SETSAll setsThe set(s) to backup. May pass in a comma-separated list of sets to back up (version 3.6.1+). Starting with asbackup 3.9.0, Aerospike Database version 5.2 or later is required for multi-set backup. Note: multi-set backup cannot be used with --filter-exp.
-B BIN1,BIN2,... or --bin-list BIN1,BIN2,...All binsThe bins to back up.
-x or --no-bins-Only backup record metadata (digest, TTL, generation count, key). WARNING: No data (bin contents) is backed up. Also, this is unrelated to the single-bin option in the Aerospike database configuration file.
-R or --no-records-Do not back up any record data (metadata or bin data). By default, asbackup includes record data, secondary index definitions, and UDF modules.
-I or --no-indexes-Do not back up any secondary index definitions.
-u or --no-udfs-Do not back up any UDF modules.
-M or --max-records N0 = all records.An approximate limit for the number of records to process. Available in server 4.9 and above. Note: this option is mutually exclusive to --partition-list and --after-digest.
-a YYYY-MM-DD_HH:MM:SS or --modified-after YYYY-MM-DD_HH:MM:SS-Back up data with last-update-time after the specified date-time. The system's local timezone applies. Available in Aerospike Database 3.12 and later. Starting with asbackup 3.9.0, Database 5.2 or later is required.
-b YYYY-MM-DD_HH:MM:SS or --modified-before YYYY-MM-DD_HH:MM:SS-Back up data with last-update-time before the specified date-cal timezone applies. Available in Aerospike Database 3.12 and later. Starting with asbackup 3.9.0, Database 5.2 or later is required.
--no-ttl-only-Include only records that have no TTL; that is, persistent records. Starting with asbackup 3.9.0, Aerospike Database 5.2 or later is required.

Partition scanning backup options

Partition list

-X, --partition-list LIST Back up a list of partition filters. Partition filters can be ranges, individual partitions, or records after a specific digest within a single partition.

This option is mutually exclusive with the -D, --after-digest option described in After specific digest, --node-list, and --max-records.

Default number of partitions to back up: 0 to 4095: all partitions.

  • LIST format: FILTER1,FILTER2,...
    • BEGIN-PARTITION: 0 to 4095.
    • Either the optional PARTITION-COUNT: 1 to 4096. Default: 1
    • Or the optional DIGEST: Base64-encoded string of desired digest to start at in specified partition.

When using multiple partition filters, each partition filter is a single scan call and cannot be parallelized with the parallel option. For more parallelizability, break up the partition filters or run a backup using only one partition filter.

When backing up only a single partition range, the range is automatically divided into parallel segments of near-equal size, each of which is backed up in parallel.


-X 361

  • Back up only partition 361

-X 361,529,841

  • Back up partitions 361, 529, and 841

-X 361-10

  • Back up 10 partitions, starting with 361 and including 370.

-X VSmeSvxNRqr46NbOqiy9gy5LTIc=

  • Back up all records after the digest VSmeSvxNRqr46NbOqiy9gy5LTIc= in its partition (which in this case is partition 2389)

-X 0-1000,2222,EjRWeJq83vEjRRI0VniavN7xI0U=

  • Back up partitions 0 to 999 (1000 partitions starting from 0)
  • Then back up partition 2222
  • Then back up all records after the digest EjRWeJq83vEjRRI0VniavN7xI0U= in its partition

After specific digest

-D , --after-digest DIGEST Back up records after the specified record digest in that record's partition and all succeeding partitions.

This option is mutually exclusive with the -X, --partition-list option described in Partition filter, --max-records, and --node-list.

  • DIGEST format: Base64-encoded string of desired digest. This is the same encoding used for backup of digests, so you can copy-and-paste digest identifiers from backup files to use as the command-line argument with -D.


-D EjRWeJq83vEjRRI0VniavN7xI0U=

Filter expression

Backups can be made of only a subset of data matching a provided Aerospike Expression. You must provide the base-64 encoding of the filter expression, which can be generated using the C client (as_exp_build_b64) or the Java client (Expression.getBytes()).

This option is mutually exclusive with multi-set backup, which is triggered by passing --set with more than one set specified.

To build an expression that filters for bin "name" = "bob", first, build the expression in the C client and print out its base 64 encoding:

as_exp_build_b64(b64_exp, as_exp_cmp_eq(as_exp_bin_str("name"), as_exp_str("bob")));
printf("%s\n", b64_exp);

This should print kwGTUQOkbmFtZaQDYm9i. Then, to run a backup with this filter expression, run:

asbackup --filter-exp kwGTUQOkbmFtZaQDYm9i ...

Backup resumption

If a backup job is interrupted, for example if you stop the backup with Ctrl-C, or it fails for any reason other than a failure to write to the disk, the backup state is saved to a .state file. Pass the path to this .state file to the --continue flag to resume the backup. All of the same command line arguments, except --remove-files, must be used when continuing a backup.

--continue STATE-FILEdisabledEnables the resumption of an interrupted backup from provided state file. All other command line arguments should match those used in the initial run (except --remove-files, which is mutually exclusive with --continue).
--state-file-dstsee belowSpecifies where to save the backup state file to. If this points to a directory, the state file is saved within the directory using the same naming convention as backup-to-directory state files. If this does not point to a directory, the path is treated as a path to the state file.

Default backup state file location

For backups to a file, the backup state is saved to a file with the same name and location as the backup file with .state appended to the filename.

For backups to a directory, the backup state is saved in the directory with name NAMESPACE.asb.state. If you supply a prefix with --output-file-prefix the prefix is used in place of NAMESPACE.

Backup to S3

To back up files to Amazon S3, prefix the file and directory names with s3://BUCKET/KEY, where BUCKET is the name of the S3 bucket to upload to or download from, and KEY is the key of the object to download/prefix of files in the S3 "directory". If using the default S3 endpoint, you must set --s3-region REGION to the region where the bucket is located. If using another endpoint, specify that endpoint with --s3-endpoint-override URL.

Files are uploaded in parts asynchronously. The maximum number of simultaneous asynchronous upload parts across all threads is controlled with --s3-max-async-uploads. Each connection to S3 is throttled, so this number may need adjustment to maximize throughput.

You can upload a maximum of 10,000 parts to S3. Each part must be between 5MB and 5GB, except for the last part which has no lower bound. When backing up to a directory, the value of --file-limit is used to calculate what part size should be used, down to the minimum size of 5MB. In other words, the part size is either --file-limit divided by 10,000, or 5MB, whichever is larger.

When backing up to a file, the estimate run before starting the backup is used in the prior calculation to find the proper upload part size. The upload part size may also be overridden with --s3-min-part-size, though it is unlikely this option will need to be used in practice.

Required S3 permissions

asbackup requires certain permissions for successful use with Amazon S3. The IAM JSON policy should include the following elements. Replace backup-bucket with the name of the S3 bucket you are using for the backup.

"Statement": [
"Action": [
"Effect": "Allow",
"Resource": [
"Action": [
"Effect": "Allow",
"Resource": [
"Version": "2012-10-17"

S3 backup options

--s3-region REGION-Sets the S3 region of the bucket being uploaded to or downloaded from. Must be set if using the default S3 endpoint.
--s3-endpoint-override URL-Sets the S3 endpoint to use. Must point to an S3-compatible storage system.
--s3-profile PROFILE_NAMEdefaultSets the S3 profile to use for credentials.
--s3-min-part-size SIZE IN MEGABYTES-An override for the minimum S3 part size to use for file uploads. By default, this size is calculated based on the expected backup file size (found either with the value of --file-limit for backup-to-directory or from the backup estimate run before backup-to-file).
--s3-max-async-downloads N32The maximum number of simultaneous download requests from S3.
--s3-max-async-uploads N16The maximum number of simultaneous upload requests from S3.
--s3-connect-timeout MILLISECONDS1000The AWS S3 client's connection timeout in milliseconds. Equivalent to cli-connect-timeout in the AWS CLI, or connectTimeoutMS in the aws-sdk-cpp client configuration.
--s3-log-level LEVELFatalThe log level of the AWS S3 C++ SDK. The possible levels are, from least to most granular.
  • Off
  • Fatal
  • Error
  • Warn
  • Info
  • Debug
  • Trace


To back up all records from a namespace test to an S3 bucket test-bucket in region us-west-1 under directory test-dir, run:

asbackup -n test -d s3://test-bucket/test-dir --s3-region us-west-1

Configuration file options

asbackup can be configured by using tools configuration files. Refer to Aerospike Tools Configuration for more details. The following options affect configuration file behavior.

--no-config-file disabledDo not read any configuration file. The configuration file options --no-config-file and only-config-file are mutually exclusive.
--instance SUFFIX-In the configuration file, you can specify a group of clusters that share a common suffix with the --instance option. Refer to Instances for more information.
--config-file PATH-Read this file after default configuration file.
--only-config-file PATH-Read only this configuration file. The configuration files options --no-config-file and only-config-file are mutually exclusive.

Secret Agent options

asbackup supports using secrets from the Aerospike Secret Agent as arguments. See Secrets for more information.

Other options

-v or --verbosedisabledOutput considerably more information about the running backup.
-m or --machine PATH-Output machine-readable status updates to the given path, typically a FIFO.
-L or --records-per-second RPS0Available only for Aerospike Database 4.7 and later.

Limit total returned records per second (RPS). If RPS is zero (the default), a records-per-second limit is not applied.

Backing up individual hosts

Use --node-list NODE1:PORT,NODE2:PORT to back up data on specific hosts. Backups will then be executed on a partition basis. PORT is the Aerospike service port, by default 3000. The --node-list flag is particularly useful when running multiple asbackup processes, for example one per Aerospike host.


If data can be retrieved from the database faster than it can be written, it may be necessary to throttle the retrieval rate. Use the --nice RATE flag to restrict the rate at which data is written. The rate is specified in MB/s.

Writing to stdout and piping

Instead of --output-file or --directory, use - to write the backup data to stdout. This is useful for pipes. The following example writes backup data to stdout with -, and pipes the output to gzip to create a compressed file:

asbackup --host HOST --namespace NAME --output-file - | gzip > FILENAME.GZ

Note that the gzip utility is single-threaded. This may cause single-CPU core saturation and create a bottleneck. To take advantage of multi-core archive utilities, consider using xz instead.

An updated method of compressing backup file data is to use the --compress runtime option. Refer to Compression and encryption.