Boto3 1 43 1

Boto3 1.43.1 — AWS SDK for Python

Overview

Boto3 is the Amazon Web Services (AWS) Software Development Kit (SDK) for Python, allowing Python developers to write software that makes use of AWS services like Amazon S3, Amazon EC2, AWS Lambda, DynamoDB, and over 200 other services. The name "Boto" refers to the freshwater dolphin native to the Amazon River.

The SDK is composed of two key packages:

• Botocore — the library providing low-level functionality shared between the Python SDK and AWS CLI
• Boto3 — the package implementing the Python SDK itself, built on top of Botocore

Boto3 provides two interfaces to AWS services:

• Low-level clients — methods that map close to 1:1 with AWS API operations, returning raw response dictionaries
• High-level resources — object-oriented abstractions over clients, providing models, collections, and waiters with more Pythonic APIs

Version Details

Version 1.43.1 requires Python 3.10 or later. Python 3.9 support has ended. It depends on:

• botocore>=1.43.1,<1.44.0
• jmespath>=0.7.1,<2.0.0
• s3transfer>=0.17.0,<0.18.0

Optional CRT dependency: botocore[crt]>=1.21.0,<2.0a0 enables AWS Common Runtime features like Amazon EventBridge Global Endpoints, S3 Multi-Region Access Points, and SigV4a signing.

What's New in 1.43.x

1.43.0 highlights:

• End of support for Python 3.9
• CloudFront cache tags — tag objects via response headers, invalidate all matching objects in single request
• Bedrock AgentCore batch evaluation with server-side orchestration, AI-powered recommendations, and AB testing
• Bedrock AgentCore configuration bundles (versioned, immutable snapshots with branch-based lineage)
• ECR removed support for registry policy V1
• GameLift Servers added DescribeContainerGroupPortMappings API for container fleets
• Transfer Family increased remote directory listing limits from 10k to 200k entries
• WorkSpaces Web added IPv6 range configuration on IP Access Settings
• AWS CRT updated to 0.32.2

1.43.1 highlights:

• Bedrock AgentCore Identity: on-behalf-of token exchange OAuth2 support
• Bedrock AgentCore Memory: metadata for LongTerm Memory Records
• DataZone: asynchronous notebook runs support
• EKS: vended logs update param for capability feature
• Kafka (MSK): ZookeeperAccess field for Client-Zookeeper connectivity control
• Observability Admin: enablement for AWS Kafka, Bedrock AgentCore Workload Identity, and OTel metric enablement
• Payment Cryptography: resource-based policies on keys (cross-account sharing), Multi-Party Approval (MPA) team association APIs
• Route 53 Global Resolver: regions support in UpdateGlobalResolver input
• SageMaker: InstancePools for flexible endpoint provisioning, Specifications for InferenceComponent per-instance-type configs
• SSO Admin: InstanceArn and IdentityStoreArn in CreateApplication/DescribeApplication responses

When to Use

Use this skill when building Python applications that integrate with AWS services:

• Cloud infrastructure management (EC2, VPC, Auto Scaling, ECS, EKS)
• Object storage operations (S3 uploads, downloads, presigned URLs, multipart transfers)
• Serverless application development (Lambda, API Gateway, DynamoDB, SQS, SNS)
• Data processing pipelines (Kinesis, Glue, EMR, Athena, Redshift)
• Identity and access management (IAM users, roles, policies, STS assume role)
• Monitoring and observability (CloudWatch metrics/logs/alarms, X-Ray)
• AI/ML workloads (Bedrock, SageMaker, Comprehend, Rekognition, Polly, Transcribe)
• Networking and CDN (Route 53, CloudFront, Global Accelerator, VPC endpoints)
• Database operations (RDS, DynamoDB, DocumentDB, Neptune, ElastiCache, Keyspaces)
• Security and compliance (KMS, GuardDuty, Config, Secrets Manager, Certificate Manager)
• Cost management and billing (Budgets, Cost Explorer, Billing Conductor)

Core Concepts

Clients vs Resources

Clients provide a low-level interface where methods map directly to AWS API operations. Parameters and responses are Python dictionaries that mirror the AWS API specification. Method names are snake-cased versions of the API operation names.

import boto3

# Low-level client
ec2 = boto3.client('ec2', region_name='us-east-1')
response = ec2.describe_instances()
for reservation in response['Reservations']:
    for instance in reservation['Instances']:
        print(instance['InstanceId'])

Resources provide a higher-level, object-oriented interface. They model AWS entities as Python objects with attributes, methods, and sub-resources.

import boto3

# High-level resource
s3 = boto3.resource('s3')
for bucket in s3.buckets.all():
    print(bucket.name)

# Access an existing bucket and list objects
bucket = s3.Bucket('my-bucket')
for obj in bucket.objects.all():
    print(obj.key, obj.size)

Resources automatically handle pagination through collections and provide waiters for common polling patterns.

Sessions

Sessions encapsulate configuration including credentials, region, and service settings. They are the factory for creating clients and resources. All clients and resources created from a session share its configuration.

import boto3

# Default session (uses default credential chain)
s3 = boto3.client('s3')

# Named session with explicit configuration
session = boto3.Session(
    profile_name='my-profile',
    region_name='eu-west-1'
)
ec2 = session.client('ec2')
dynamodb = session.resource('dynamodb')

Collections

Collections provide an iterable interface to groups of resources. They support filtering, limiting, and batch actions without making API calls until iteration begins.

# Filter EC2 instances by state
running = ec2.instances.filter(Filters=[{'Name': 'instance-state-name', 'Values': ['running']}])
for instance in running:
    print(instance.id, instance.state['Name'])

# Batch terminate instances
instances_to_stop = [i for i in running if i.tags and any(t['Key'] == 'Environment' and t['Value'] == 'dev' for t in i.tags)]
ec2.instances.filter(InstanceIds=[i.id for i in instances_to_stop]).stop()

Collections are chainable and return copies, supporting lazy evaluation patterns.

Paginators

Paginators automatically handle pagination across AWS API responses that return partial results. They abstract away NextToken, Marker, and other pagination parameters.

# List all S3 objects (handles pagination automatically)
paginator = s3_client.get_paginator('list_objects_v2')
for page in paginator.paginate(Bucket='my-bucket'):
    for obj in page.get('Contents', []):
        print(obj['Key'])

# Pass pagination config to limit page size
for page in paginator.paginate(Bucket='my-bucket', PaginationConfig={'MaxItems': 100, 'PageSize': 10}):
    for obj in page.get('Contents', []):
        print(obj['Key'])

Waiters

Waiters poll an AWS resource until it reaches a desired state or times out. They use exponential backoff internally.

# Wait for EC2 instance to be running
instance = ec2.create_instances(ImageId='ami-12345', MinCount=1, MaxCount=1)[0]
instance.wait_until_running()

# Wait for S3 bucket to exist
s3_client = boto3.client('s3')
s3_client.create_bucket(Bucket='my-bucket')
s3_client.get_waiter('bucket_exists').wait(Bucket='my-bucket')

Available waiters can be listed via client.waiter_names.

Advanced Topics

Credentials and Authentication: Credential chain, profiles, IAM roles, SSO, web identity, and environment variables → Credentials and Authentication

Configuration and Client Options: Config objects, retry modes, proxies, timeouts, dual-stack, FIPS endpoints, and client context parameters → Configuration and Client Options

S3 Operations: Managed transfers, multipart uploads, presigned URLs, transfer configuration, S3 control, and S3 Express → S3 Operations

Error Handling and Retries: Botocore exceptions, service exceptions, retry modes (legacy/standard/adaptive), and error parsing → Error Handling and Retries

Common Service Patterns: EC2 instance management, Lambda deployment, DynamoDB operations, SQS messaging, CloudWatch monitoring, and IAM administration → Common Service Patterns

Advanced Patterns: Async clients, custom events, extensibility hooks, STS assume role, and cross-account access → Advanced Patterns