Python and ElastiCache - Amazon ElastiCache
Tutorial PrerequisitesTutorial: Creating ElastiCache clusters and usersTutorial: Connecting to ElastiCacheUsage examples

Python and ElastiCache

In this tutorial, you use the AWS SDK for Python (Boto3) to write simple programs to perform the following ElastiCache (Redis OSS) operations:

  • Create ElastiCache (Redis OSS) clusters (cluster mode enabled and cluster mode disabled)

  • Check if users or user groups exist, otherwise create them. (This feature is availably with Valkey 7.2 and onwards, and with Redis OSS 6.0 onwards.)

  • Connect to ElastiCache

  • Perform operations such as setting and getting strings, reading from and writing to steams and publishing and subscribing from Pub/Sub channel.

As you work through this tutorial, you can refer to the AWS SDK for Python (Boto) documentation. The following section is specific to ElastiCache: ElastiCache low-level client

Tutorial Prerequisites

Tutorial: Creating ElastiCache clusters and users

The following examples use the boto3 SDK for ElastiCache (Redis OSS) management operations (cluster or user creation) and redis-py/redis-py-cluster for data handling.

Create a cluster mode disabled cluster

Copy the following program and paste it into a file named CreateClusterModeDisabledCluster.py.

import boto3
import logging

logging.basicConfig(level=logging.INFO)
client = boto3.client('elasticache')

def create_cluster_mode_disabled(CacheNodeType='cache.t3.small',EngineVersion='6.0',NumCacheClusters=2,ReplicationGroupDescription='Sample cache cluster',ReplicationGroupId=None):
    """Creates an ElastiCache Cluster with cluster mode disabled

    Returns a dictionary with the API response

    :param CacheNodeType: Node type used on the cluster. If not specified, cache.t3.small will be used
    Refer to https://docs.aws.amazon.com/AmazonElastiCache/latest/dg/CacheNodes.SupportedTypes.html for supported node types
    :param EngineVersion: Engine version to be used. If not specified, latest will be used.
    :param NumCacheClusters: Number of nodes in the cluster. Minimum 1 (just a primary node) and maximun 6 (1 primary and 5 replicas).
    If not specified, cluster will be created with 1 primary and 1 replica.
    :param ReplicationGroupDescription: Description for the cluster.
    :param ReplicationGroupId: Name for the cluster
    :return: dictionary with the API results

    """
    if not ReplicationGroupId:
        return 'ReplicationGroupId parameter is required'

    response = client.create_replication_group(
        AutomaticFailoverEnabled=True,
        CacheNodeType=CacheNodeType,
        Engine='valkey',
        EngineVersion=EngineVersion,
        NumCacheClusters=NumCacheClusters,
        ReplicationGroupDescription=ReplicationGroupDescription,
        ReplicationGroupId=ReplicationGroupId,
        SnapshotRetentionLimit=30,
    )
    return response


if __name__ == '__main__':
    
    # Creates an ElastiCache Cluster mode disabled cluster, based on cache.m6g.large nodes, Valkey 7.2, one primary and two replicas
    elasticacheResponse = create_cluster_mode_disabled(
        #CacheNodeType='cache.m6g.large', 
        EngineVersion='7.2',
        NumCacheClusters=3,
        ReplicationGroupDescription='Valkey cluster mode disabled with replicas',
        ReplicationGroupId='valkey202104053'
        )
    
    logging.info(elasticacheResponse)

To run the program, enter the following command:

python CreateClusterModeDisabledCluster.py

For more information, see Managing clusters in ElastiCache.

Create a cluster mode disabled cluster with TLS and RBAC

To ensure security, you can use Transport Layer Security (TLS) and Role-Based Access Control (RBAC) when creating a cluster mode disabled cluster. Unlike Valkey or Redis OSS AUTH, where all authenticated clients have full replication group access if their token is authenticated, RBAC enables you to control cluster access through user groups. These user groups are designed as a way to organize access to replication groups. For more information, see Role-Based Access Control (RBAC).

Copy the following program and paste it into a file named ClusterModeDisabledWithRBAC.py.

import boto3
import logging

logging.basicConfig(level=logging.INFO)
client = boto3.client('elasticache')

def create_cluster_mode_disabled_rbac(CacheNodeType='cache.t3.small',EngineVersion='6.0',NumCacheClusters=2,ReplicationGroupDescription='Sample cache cluster',ReplicationGroupId=None, UserGroupIds=None, SecurityGroupIds=None,CacheSubnetGroupName=None):
    """Creates an ElastiCache Cluster with cluster mode disabled and RBAC

    Returns a dictionary with the API response

    :param CacheNodeType: Node type used on the cluster. If not specified, cache.t3.small will be used
    Refer to https://docs.aws.amazon.com/AmazonElastiCache/latest/dg/CacheNodes.SupportedTypes.html for supported node types
    :param EngineVersion: Engine version to be used. If not specified, latest will be used.
    :param NumCacheClusters: Number of nodes in the cluster. Minimum 1 (just a primary node) and maximun 6 (1 primary and 5 replicas).
    If not specified, cluster will be created with 1 primary and 1 replica.
    :param ReplicationGroupDescription: Description for the cluster.
    :param ReplicationGroupId: Mandatory name for the cluster.
    :param UserGroupIds: The ID of the user group to be assigned to the cluster.
    :param SecurityGroupIds: List of security groups to be assigned. If not defined, default will be used
    :param CacheSubnetGroupName: subnet group where the cluster will be placed. If not defined, default will be used.
    :return: dictionary with the API results

    """
    if not ReplicationGroupId:
        return {'Error': 'ReplicationGroupId parameter is required'}
    elif not isinstance(UserGroupIds,(list)):
        return {'Error': 'UserGroupIds parameter is required and must be a list'}

    params={'AutomaticFailoverEnabled': True, 
            'CacheNodeType': CacheNodeType, 
            'Engine': 'valkey', 
            'EngineVersion': EngineVersion, 
            'NumCacheClusters': NumCacheClusters, 
            'ReplicationGroupDescription': ReplicationGroupDescription, 
            'ReplicationGroupId': ReplicationGroupId, 
            'SnapshotRetentionLimit': 30, 
            'TransitEncryptionEnabled': True, 
            'UserGroupIds':UserGroupIds
        }

    # defaults will be used if CacheSubnetGroupName or SecurityGroups are not explicit.
    if isinstance(SecurityGroupIds,(list)):
        params.update({'SecurityGroupIds':SecurityGroupIds})
    if CacheSubnetGroupName:
        params.update({'CacheSubnetGroupName':CacheSubnetGroupName})

    response = client.create_replication_group(**params)
    return response

if __name__ == '__main__':

    # Creates an ElastiCache Cluster mode disabled cluster, based on cache.m6g.large nodes, Valkey 7.2, one primary and two replicas.
    # Assigns the existent user group "mygroup" for RBAC authentication
   
    response=create_cluster_mode_disabled_rbac(
        CacheNodeType='cache.m6g.large',
        EngineVersion='7.2',
        NumCacheClusters=3,
        ReplicationGroupDescription='Valkey cluster mode disabled with replicas',
        ReplicationGroupId='valkey202104',
        UserGroupIds=[
            'mygroup'
        ],
        SecurityGroupIds=[
            'sg-7cc73803'
        ],
        CacheSubnetGroupName='default'
    )

    logging.info(response)

To run the program, enter the following command:

python ClusterModeDisabledWithRBAC.py

For more information, see Managing clusters in ElastiCache.

Create a cluster mode enabled cluster

Copy the following program and paste it into a file named ClusterModeEnabled.py.

import boto3
import logging

logging.basicConfig(level=logging.INFO)
client = boto3.client('elasticache')

def create_cluster_mode_enabled(CacheNodeType='cache.t3.small',EngineVersion='6.0',NumNodeGroups=1,ReplicasPerNodeGroup=1, ReplicationGroupDescription='Sample cache with cluster mode enabled',ReplicationGroupId=None):
    """Creates an ElastiCache Cluster with cluster mode enabled

    Returns a dictionary with the API response

    :param CacheNodeType: Node type used on the cluster. If not specified, cache.t3.small will be used
    Refer to https://docs.aws.amazon.com/AmazonElastiCache/latest/dg/CacheNodes.SupportedTypes.html for supported node types
    :param EngineVersion: Engine version to be used. If not specified, latest will be used.
    :param NumNodeGroups: Number of shards in the cluster. Minimum 1 and maximun 90.
    If not specified, cluster will be created with 1 shard.
    :param ReplicasPerNodeGroup: Number of replicas per shard. If not specified 1 replica per shard will be created.
    :param ReplicationGroupDescription: Description for the cluster.
    :param ReplicationGroupId: Name for the cluster
    :return: dictionary with the API results

    """
    if not ReplicationGroupId:
        return 'ReplicationGroupId parameter is required'
    
    response = client.create_replication_group(
        AutomaticFailoverEnabled=True,
        CacheNodeType=CacheNodeType,
        Engine='valkey',
        EngineVersion=EngineVersion,
        ReplicationGroupDescription=ReplicationGroupDescription,
        ReplicationGroupId=ReplicationGroupId,
    #   Creates a cluster mode enabled cluster with 1 shard(NumNodeGroups), 1 primary node (implicit) and 2 replicas (replicasPerNodeGroup)
        NumNodeGroups=NumNodeGroups,
        ReplicasPerNodeGroup=ReplicasPerNodeGroup,
        CacheParameterGroupName='default.valkey7.2.cluster.on'
    )

    return response


# Creates a cluster mode enabled 
response = create_cluster_mode_enabled(
    CacheNodeType='cache.m6g.large',
    EngineVersion='6.0',
    ReplicationGroupDescription='Valkey cluster mode enabled with replicas',
    ReplicationGroupId='valkey20210',
#   Creates a cluster mode enabled cluster with 1 shard(NumNodeGroups), 1 primary (implicit) and 2 replicas (replicasPerNodeGroup)
    NumNodeGroups=2,
    ReplicasPerNodeGroup=1,
)

logging.info(response)

To run the program, enter the following command:

python ClusterModeEnabled.py

For more information, see Managing clusters in ElastiCache.

Create a cluster mode enabled cluster with TLS and RBAC

To ensure security, you can use Transport Layer Security (TLS) and Role-Based Access Control (RBAC) when creating a cluster mode enabled cluster. Unlike Valkey or Redis OSS AUTH, where all authenticated clients have full replication group access if their token is authenticated, RBAC enables you to control cluster access through user groups. These user groups are designed as a way to organize access to replication groups. For more information, see Role-Based Access Control (RBAC).

Copy the following program and paste it into a file named ClusterModeEnabledWithRBAC.py.

import boto3
import logging

logging.basicConfig(level=logging.INFO)
client = boto3.client('elasticache')

def create_cluster_mode_enabled(CacheNodeType='cache.t3.small',EngineVersion='6.0',NumNodeGroups=1,ReplicasPerNodeGroup=1, ReplicationGroupDescription='Sample cache with cluster mode enabled',ReplicationGroupId=None,UserGroupIds=None, SecurityGroupIds=None,CacheSubnetGroupName=None,CacheParameterGroupName='default.valkey7.2.cluster.on'):
    """Creates an ElastiCache Cluster with cluster mode enabled and RBAC

    Returns a dictionary with the API response

    :param CacheNodeType: Node type used on the cluster. If not specified, cache.t3.small will be used
    Refer to https://docs.aws.amazon.com/AmazonElastiCache/latest/dg/CacheNodes.SupportedTypes.html for supported node types
    :param EngineVersion: Engine version to be used. If not specified, latest will be used.
    :param NumNodeGroups: Number of shards in the cluster. Minimum 1 and maximun 90.
    If not specified, cluster will be created with 1 shard.
    :param ReplicasPerNodeGroup: Number of replicas per shard. If not specified 1 replica per shard will be created.
    :param ReplicationGroupDescription: Description for the cluster.
    :param ReplicationGroupId: Name for the cluster.
    :param CacheParameterGroupName: Parameter group to be used. Must be compatible with the engine version and cluster mode enabled.
    :return: dictionary with the API results

    """
    if not ReplicationGroupId:
        return 'ReplicationGroupId parameter is required'
    elif not isinstance(UserGroupIds,(list)):
        return {'Error': 'UserGroupIds parameter is required and must be a list'}

    params={'AutomaticFailoverEnabled': True, 
            'CacheNodeType': CacheNodeType, 
            'Engine': 'valkey', 
            'EngineVersion': EngineVersion, 
            'ReplicationGroupDescription': ReplicationGroupDescription, 
            'ReplicationGroupId': ReplicationGroupId, 
            'SnapshotRetentionLimit': 30, 
            'TransitEncryptionEnabled': True, 
            'UserGroupIds':UserGroupIds,
            'NumNodeGroups': NumNodeGroups,
            'ReplicasPerNodeGroup': ReplicasPerNodeGroup,
            'CacheParameterGroupName': CacheParameterGroupName
        }

    # defaults will be used if CacheSubnetGroupName or SecurityGroups are not explicit.
    if isinstance(SecurityGroupIds,(list)):
        params.update({'SecurityGroupIds':SecurityGroupIds})
    if CacheSubnetGroupName:
        params.update({'CacheSubnetGroupName':CacheSubnetGroupName})

    response = client.create_replication_group(**params)
    return response

if __name__ == '__main__':
    # Creates a cluster mode enabled cluster
    response = create_cluster_mode_enabled(
        CacheNodeType='cache.m6g.large',
        EngineVersion='7.2',
        ReplicationGroupDescription='Valkey cluster mode enabled with replicas',
        ReplicationGroupId='valkey2021',
    #   Creates a cluster mode enabled cluster with 1 shard(NumNodeGroups), 1 primary (implicit) and 2 replicas (replicasPerNodeGroup)
        NumNodeGroups=2,
        ReplicasPerNodeGroup=1,
        UserGroupIds=[
            'mygroup'
        ],
        SecurityGroupIds=[
            'sg-7cc73803'
        ],
        CacheSubnetGroupName='default'

    )
    
    logging.info(response)

To run the program, enter the following command:

python ClusterModeEnabledWithRBAC.py

For more information, see Managing clusters in ElastiCache.

Check if users/usergroup exists, otherwise create them

With RBAC, you create users and assign them specific permissions by using an access string. You assign the users to user groups aligned with a specific role (administrators, human resources) that are then deployed to one or more ElastiCache (Redis OSS) replication groups. By doing this, you can establish security boundaries between clients using the same Valkey or Redis OSS replication group or groups and prevent clients from accessing each other’s data. For more information, see Role-Based Access Control (RBAC).

Copy the following program and paste it into a file named UserAndUserGroups.py. Update the mechanism for supplying credentials. Credentials in this example are shown as replaceable and assigned an undeclared item. Avoid hard-coding credentials.

This example uses an access string with the permissions for the user. For more information on access strings see Specifying Permissions Using an Access String.

import boto3
import logging

logging.basicConfig(level=logging.INFO)
client = boto3.client('elasticache')

def check_user_exists(UserId):
    """Checks if UserId exists

    Returns True if UserId exists, otherwise False
    :param UserId: ElastiCache User ID
    :return: True|False
    """
    try:
        response = client.describe_users(
            UserId=UserId,
        )
        if response['Users'][0]['UserId'].lower() == UserId.lower():
            return True
    except Exception as e:
        if e.response['Error']['Code'] == 'UserNotFound':
            logging.info(e.response['Error'])
            return False
        else:
            raise

def check_group_exists(UserGroupId):
    """Checks if UserGroupID exists

    Returns True if Group ID exists, otherwise False
    :param UserGroupId: ElastiCache User ID
    :return: True|False
    """

    try:
        response = client.describe_user_groups(
            UserGroupId=UserGroupId
        )
        if response['UserGroups'][0]['UserGroupId'].lower() == UserGroupId.lower():
            return True
    except Exception as e:
        if e.response['Error']['Code'] == 'UserGroupNotFound':
            logging.info(e.response['Error'])
            return False
        else:
            raise

def create_user(UserId=None,UserName=None,Password=None,AccessString=None):
    """Creates a new user

    Returns the ARN for the newly created user or the error message
    :param UserId: ElastiCache user ID. User IDs must be unique
    :param UserName: ElastiCache user name. ElastiCache allows multiple users with the same name as long as the associated user ID is unique.
    :param Password: Password for user. Must have at least 16 chars.
    :param AccessString: Access string with the permissions for the user. 
    :return: user ARN
    """
    try:
        response = client.create_user(
            UserId=UserId,
            UserName=UserName,
            Engine='Redis',
            Passwords=[Password],
            AccessString=AccessString,
            NoPasswordRequired=False
        )
        return response['ARN']
    except Exception as e:
        logging.info(e.response['Error'])
        return e.response['Error']

def create_group(UserGroupId=None, UserIds=None):
    """Creates a new group.
    A default user is required (mandatory) and should be specified in the UserIds list

    Return: Group ARN
    :param UserIds: List with user IDs to be associated with the new group. A default user is required
    :param UserGroupId: The ID (name) for the group
    :return: Group ARN
    """
    try:
        response = client.create_user_group(
            UserGroupId=UserGroupId,
            Engine='Redis',
            UserIds=UserIds
        )
        return response['ARN']
    except Exception as e:
        logging.info(e.response['Error'])


if __name__ == '__main__':
    
    groupName='mygroup2'
    userName = 'myuser2'
    userId=groupName+'-'+userName

    # Creates a new user if the user ID does not exist.
    for tmpUserId,tmpUserName in [ (userId,userName), (groupName+'-default','default')]:
        if not check_user_exists(tmpUserId):
            response=create_user(UserId=tmpUserId, UserName=EXAMPLE,Password=EXAMPLE,AccessString='on ~* +@all')
            logging.info(response)
        # assigns the new user ID to the user group
    if not check_group_exists(groupName):
        UserIds = [ userId , groupName+'-default']
        response=create_group(UserGroupId=groupName,UserIds=UserIds)
        logging.info(response)

To run the program, enter the following command:

python UserAndUserGroups.py

Tutorial: Connecting to ElastiCache

The following examples use the Valkey or Redis OSS client to connect to ElastiCache.

Connecting to a cluster mode disabled cluster

Copy the following program and paste it into a file named ConnectClusterModeDisabled.py. Update the mechanism for supplying credentials. Credentials in this example are shown as replaceable and assigned an undeclared item. Avoid hard-coding credentials.

from redis import Redis
import logging

logging.basicConfig(level=logging.INFO)
redis = Redis(host='primary.xxx.yyyyyy.zzz1.cache.amazonaws.com', port=6379, decode_responses=True, ssl=True, username=example, password=EXAMPLE)

if redis.ping():
    logging.info("Connected to Redis")

To run the program, enter the following command:

python ConnectClusterModeDisabled.py

Connecting to a cluster mode enabled cluster

Copy the following program and paste it into a file named ConnectClusterModeEnabled.py.

from rediscluster import RedisCluster
import logging

logging.basicConfig(level=logging.INFO)
redis = RedisCluster(startup_nodes=[{"host": "xxx.yyy.clustercfg.zzz1.cache.amazonaws.com","port": "6379"}], decode_responses=True,skip_full_coverage_check=True)

if redis.ping():
    logging.info("Connected to Redis")

To run the program, enter the following command:

python ConnectClusterModeEnabled.py

Usage examples

The following examples use the boto3 SDK for ElastiCache to work with ElastiCache (Redis OSS).

Set and Get strings

Copy the following program and paste it into a file named SetAndGetStrings.py.

import time
import logging
logging.basicConfig(level=logging.INFO,format='%(asctime)s: %(message)s')

keyName='mykey'
currTime=time.ctime(time.time())

# Set the key 'mykey' with the current date and time as value. 
# The Key will expire and removed from cache in 60 seconds.
redis.set(keyName, currTime, ex=60)

# Sleep just for better illustration of TTL (expiration) value
time.sleep(5)

# Retrieve the key value and current TTL
keyValue=redis.get(keyName)
keyTTL=redis.ttl(keyName)

logging.info("Key {} was set at {} and has {} seconds until expired".format(keyName, keyValue, keyTTL))

To run the program, enter the following command:

python SetAndGetStrings.py

Set and Get a hash with multiple items

Copy the following program and paste it into a file named SetAndGetHash.py.

import logging
import time

logging.basicConfig(level=logging.INFO,format='%(asctime)s: %(message)s')

keyName='mykey'
keyValues={'datetime': time.ctime(time.time()), 'epochtime': time.time()}

# Set the hash 'mykey' with the current date and time in human readable format (datetime field) and epoch number (epochtime field). 
redis.hset(keyName, mapping=keyValues)

# Set the key to expire and removed from cache in 60 seconds.
redis.expire(keyName, 60)

# Sleep just for better illustration of TTL (expiration) value
time.sleep(5)

# Retrieves all the fields and current TTL
keyValues=redis.hgetall(keyName)
keyTTL=redis.ttl(keyName)

logging.info("Key {} was set at {} and has {} seconds until expired".format(keyName, keyValues, keyTTL))

To run the program, enter the following command:

python SetAndGetHash.py

Publish (write) and subscribe (read) from a Pub/Sub channel

Copy the following program and paste it into a file named PubAndSub.py.

import logging
import time

def handlerFunction(message):
    """Prints message got from PubSub channel to the log output

    Return None
    :param message: message to log
    """
    logging.info(message)

logging.basicConfig(level=logging.INFO)
redis = Redis(host="redis202104053.tihewd.ng.0001.use1.cache.amazonaws.com", port=6379, decode_responses=True)


# Creates the subscriber connection on "mychannel"
subscriber = redis.pubsub()
subscriber.subscribe(**{'mychannel': handlerFunction})

# Creates a new thread to watch for messages while the main process continues with its routines
thread = subscriber.run_in_thread(sleep_time=0.01)

# Creates publisher connection on "mychannel"
redis.publish('mychannel', 'My message')

# Publishes several messages. Subscriber thread will read and print on log.
while True:
    redis.publish('mychannel',time.ctime(time.time()))
    time.sleep(1)

To run the program, enter the following command:

python PubAndSub.py

Write and read from a stream

Copy the following program and paste it into a file named ReadWriteStream.py.

from redis import Redis
import redis.exceptions as exceptions
import logging
import time
import threading

logging.basicConfig(level=logging.INFO)

def writeMessage(streamName):
    """Starts a loop writting the current time and thread name to 'streamName'

    :param streamName: Stream (key) name to write messages.
    """
    fieldsDict={'writerId':threading.currentThread().getName(),'myvalue':None}
    while True:
        fieldsDict['myvalue'] = time.ctime(time.time())
        redis.xadd(streamName,fieldsDict)
        time.sleep(1)

def readMessage(groupName=None,streamName=None):
    """Starts a loop reading from 'streamName'
    Multiple threads will read from the same stream consumer group. Consumer group is used to coordinate data distribution.
    Once a thread acknowleges the message, it won't be provided again. If message wasn't acknowledged, it can be served to another thread.

    :param groupName: stream group were multiple threads will read.
    :param streamName: Stream (key) name where messages will be read.
    """

    readerID=threading.currentThread().getName()
    while True:
        try:
            # Check if the stream has any message
            if redis.xlen(streamName)>0:
                # Check if if the messages are new (not acknowledged) or not (already processed)
                streamData=redis.xreadgroup(groupName,readerID,{streamName:'>'},count=1)
                if len(streamData) > 0:
                    msgId,message = streamData[0][1][0]
                    logging.info("{}: Got {} from ID {}".format(readerID,message,msgId))
                    #Do some processing here. If the message has been processed sucessfuly, acknowledge it and (optional) delete the message.
                    redis.xack(streamName,groupName,msgId)
                    logging.info("Stream message ID {} read and processed successfuly by {}".format(msgId,readerID))
                    redis.xdel(streamName,msgId)
            else:
                pass
        except:
            raise
            
        time.sleep(0.5)

# Creates the stream 'mystream' and consumer group 'myworkergroup' where multiple threads will write/read.
try:
    redis.xgroup_create('mystream','myworkergroup',mkstream=True)
except exceptions.ResponseError as e:
    logging.info("Consumer group already exists. Will continue despite the error: {}".format(e))
except:
    raise

# Starts 5 writer threads.
for writer_no in range(5):
    writerThread = threading.Thread(target=writeMessage, name='writer-'+str(writer_no), args=('mystream',),daemon=True)
    writerThread.start()

# Starts 10 reader threads
for reader_no in range(10):
    readerThread = threading.Thread(target=readMessage, name='reader-'+str(reader_no), args=('myworkergroup','mystream',),daemon=True)
    readerThread.daemon = True
    readerThread.start()

# Keep the code running for 30 seconds
time.sleep(30)

To run the program, enter the following command:

python ReadWriteStream.py