DynamoDB

Overview

DevCloud DynamoDB persists tables and items in SQLite (dynamodb.db under the service’s data_dir), with table metadata also held in an in-memory index guarded by an RWMutex.

All ten AttributeValue types are supported: S, N, B, BOOL, NULL, L (List), M (Map), and the SS / NS / BS sets.

Supported APIs

These 29 operations are hand-verified, implemented by the provider, not by the CRUD engine. Everything else DynamoDB models is served at a lower tier or not at all; fidelity-manifest.md is the per-operation answer.

Operation	Description
CreateTable	Create table with partition key (HASH), optional sort key (RANGE), GSIs, LSIs and a StreamSpecification
UpdateTable / DescribeTable / DeleteTable / ListTables	Manage and inspect tables
PutItem	Insert or overwrite an item; honours `ConditionExpression`
GetItem	Retrieve an item by primary key; honours `ProjectionExpression`
UpdateItem	Apply an `UpdateExpression` to one item
DeleteItem	Delete an item by primary key; honours `ConditionExpression`
Query	Query by partition key, against the table or a named `IndexName`; honours `FilterExpression`
Scan	Full table scan; honours `FilterExpression`
BatchGetItem / BatchWriteItem	Multi-item reads and writes
TransactGetItems / TransactWriteItems	Transactional reads and writes
UpdateTimeToLive / DescribeTimeToLive	Store and read back a table’s TTL configuration
TagResource / UntagResource / ListTagsOfResource	Manage table tags

Writes to a table created with StreamEnabled are published to the dynamodbstreams service, which is what makes DynamoDB Streams -> Lambda event source mappings fire.

boto3 Examples

Create a table and write items

import boto3

dynamodb = boto3.client(
    "dynamodb",
    endpoint_url="http://localhost:4747",
    aws_access_key_id="test",
    aws_secret_access_key="test",
    region_name="us-east-1",
)

# Create table
dynamodb.create_table(
    TableName="users",
    KeySchema=[
        {"AttributeName": "pk", "KeyType": "HASH"},
        {"AttributeName": "sk", "KeyType": "RANGE"},
    ],
    AttributeDefinitions=[
        {"AttributeName": "pk", "AttributeType": "S"},
        {"AttributeName": "sk", "AttributeType": "S"},
    ],
    BillingMode="PAY_PER_REQUEST",
)

# Put item
dynamodb.put_item(
    TableName="users",
    Item={
        "pk": {"S": "user#123"},
        "sk": {"S": "profile"},
        "name": {"S": "Alice"},
        "age": {"N": "30"},
    },
)

# Get item
response = dynamodb.get_item(
    TableName="users",
    Key={"pk": {"S": "user#123"}, "sk": {"S": "profile"}},
)
print(response["Item"])

Query and scan

# Query by partition key
response = dynamodb.query(
    TableName="users",
    KeyConditionExpression="pk = :pk",
    ExpressionAttributeValues={":pk": {"S": "user#123"}},
)
for item in response["Items"]:
    print(item)

# Full table scan
response = dynamodb.scan(TableName="users")
for item in response["Items"]:
    print(item)

AWS CLI Examples

# Create table
aws --endpoint-url http://localhost:4747 dynamodb create-table \
  --table-name my-table \
  --key-schema AttributeName=pk,KeyType=HASH \
  --attribute-definitions AttributeName=pk,AttributeType=S \
  --billing-mode PAY_PER_REQUEST

# Put item
aws --endpoint-url http://localhost:4747 dynamodb put-item \
  --table-name my-table \
  --item '{"pk": {"S": "key1"}, "data": {"S": "value1"}}'

# Get item
aws --endpoint-url http://localhost:4747 dynamodb get-item \
  --table-name my-table \
  --key '{"pk": {"S": "key1"}}'

Known Limitations

TTL is configuration only. UpdateTimeToLive stores the attribute name and echoes it back; nothing sweeps expired items, so a row past its TTL is still returned.
No pagination or capacity reporting. Responses carry no LastEvaluatedKey and no ConsumedCapacity, so Query/Scan return the whole matching set in one page and code that loops on the cursor sees one iteration.
PartiQL subset only: ExecuteStatement, ExecuteTransaction and BatchExecuteStatement support single-table SELECT, INSERT, UPDATE, and DELETE. WHERE clauses are limited to simple equality (=). No nested paths, subqueries, or arithmetic.
Backups and global tables answer from the CRUD engine. CreateBackup, CreateGlobalTable, DescribeContinuousBackups and their neighbours return stored, plausible shapes with no behaviour behind them. However, ExportTableToPointInTime, ImportTable, RestoreTableFromBackup, and RestoreTableToPointInTime are fully handled by the provider (cloning tables and items in SQLite).
No Kinesis active replication. EnableKinesisStreamingDestination and DisableKinesisStreamingDestination persist configuration, but write events are not actively replicated to Kinesis streams.
No provisioned-throughput accounting or throttling; BillingMode is recorded, never enforced
Single account model (account ID: 000000000000)