使 DAX DynamoDB 加速讀取 AWS SDK

以下程式碼範例顯示做法：

• 使用和用SDK戶端建立資料DAX並將資料寫入資料表。

• 使用兩個用戶端取得、查詢和掃描表格，並比較其效能。

如需詳細資訊，請參閱使用 DynamoDB Accelerator 用戶端開發。

Python

SDK對於 Python（肉毒桿菌 3）

注意

還有更多關於 GitHub。尋找完整範例，並了解如何在 AWS 程式碼範例儲存庫中設定和執行。

使用DAX或 Boto3 用戶端建立資料表。

import boto3


def create_dax_table(dyn_resource=None):
    """
    Creates a DynamoDB table.

    :param dyn_resource: Either a Boto3 or DAX resource.
    :return: The newly created table.
    """
    if dyn_resource is None:
        dyn_resource = boto3.resource("dynamodb")

    table_name = "TryDaxTable"
    params = {
        "TableName": table_name,
        "KeySchema": [
            {"AttributeName": "partition_key", "KeyType": "HASH"},
            {"AttributeName": "sort_key", "KeyType": "RANGE"},
        ],
        "AttributeDefinitions": [
            {"AttributeName": "partition_key", "AttributeType": "N"},
            {"AttributeName": "sort_key", "AttributeType": "N"},
        ],
        "ProvisionedThroughput": {"ReadCapacityUnits": 10, "WriteCapacityUnits": 10},
    }
    table = dyn_resource.create_table(**params)
    print(f"Creating {table_name}...")
    table.wait_until_exists()
    return table


if __name__ == "__main__":
    dax_table = create_dax_table()
    print(f"Created table.")

將測試資料寫入資料表。

import boto3


def write_data_to_dax_table(key_count, item_size, dyn_resource=None):
    """
    Writes test data to the demonstration table.

    :param key_count: The number of partition and sort keys to use to populate the
                      table. The total number of items is key_count * key_count.
    :param item_size: The size of non-key data for each test item.
    :param dyn_resource: Either a Boto3 or DAX resource.
    """
    if dyn_resource is None:
        dyn_resource = boto3.resource("dynamodb")

    table = dyn_resource.Table("TryDaxTable")
    some_data = "X" * item_size

    for partition_key in range(1, key_count + 1):
        for sort_key in range(1, key_count + 1):
            table.put_item(
                Item={
                    "partition_key": partition_key,
                    "sort_key": sort_key,
                    "some_data": some_data,
                }
            )
            print(f"Put item ({partition_key}, {sort_key}) succeeded.")


if __name__ == "__main__":
    write_key_count = 10
    write_item_size = 1000
    print(
        f"Writing {write_key_count*write_key_count} items to the table. "
        f"Each item is {write_item_size} characters."
    )
    write_data_to_dax_table(write_key_count, write_item_size)

為客戶端和 Boto3 DAX 客戶端獲取多個迭代項目，並報告每個客戶端花費的時間。

import argparse
import sys
import time
import amazondax
import boto3


def get_item_test(key_count, iterations, dyn_resource=None):
    """
    Gets items from the table a specified number of times. The time before the
    first iteration and the time after the last iteration are both captured
    and reported.

    :param key_count: The number of items to get from the table in each iteration.
    :param iterations: The number of iterations to run.
    :param dyn_resource: Either a Boto3 or DAX resource.
    :return: The start and end times of the test.
    """
    if dyn_resource is None:
        dyn_resource = boto3.resource("dynamodb")

    table = dyn_resource.Table("TryDaxTable")
    start = time.perf_counter()
    for _ in range(iterations):
        for partition_key in range(1, key_count + 1):
            for sort_key in range(1, key_count + 1):
                table.get_item(
                    Key={"partition_key": partition_key, "sort_key": sort_key}
                )
                print(".", end="")
                sys.stdout.flush()
    print()
    end = time.perf_counter()
    return start, end


if __name__ == "__main__":
    # pylint: disable=not-context-manager
    parser = argparse.ArgumentParser()
    parser.add_argument(
        "endpoint_url",
        nargs="?",
        help="When specified, the DAX cluster endpoint. Otherwise, DAX is not used.",
    )
    args = parser.parse_args()

    test_key_count = 10
    test_iterations = 50
    if args.endpoint_url:
        print(
            f"Getting each item from the table {test_iterations} times, "
            f"using the DAX client."
        )
        # Use a with statement so the DAX client closes the cluster after completion.
        with amazondax.AmazonDaxClient.resource(endpoint_url=args.endpoint_url) as dax:
            test_start, test_end = get_item_test(
                test_key_count, test_iterations, dyn_resource=dax
            )
    else:
        print(
            f"Getting each item from the table {test_iterations} times, "
            f"using the Boto3 client."
        )
        test_start, test_end = get_item_test(test_key_count, test_iterations)
    print(
        f"Total time: {test_end - test_start:.4f} sec. Average time: "
        f"{(test_end - test_start)/ test_iterations}."
    )

在表格中查詢用戶端和 Boto3 DAX 用戶端的多次迭代，並報告每個用戶端所花費的時間。

import argparse
import time
import sys
import amazondax
import boto3
from boto3.dynamodb.conditions import Key


def query_test(partition_key, sort_keys, iterations, dyn_resource=None):
    """
    Queries the table a specified number of times. The time before the
    first iteration and the time after the last iteration are both captured
    and reported.

    :param partition_key: The partition key value to use in the query. The query
                          returns items that have partition keys equal to this value.
    :param sort_keys: The range of sort key values for the query. The query returns
                      items that have sort key values between these two values.
    :param iterations: The number of iterations to run.
    :param dyn_resource: Either a Boto3 or DAX resource.
    :return: The start and end times of the test.
    """
    if dyn_resource is None:
        dyn_resource = boto3.resource("dynamodb")

    table = dyn_resource.Table("TryDaxTable")
    key_condition_expression = Key("partition_key").eq(partition_key) & Key(
        "sort_key"
    ).between(*sort_keys)

    start = time.perf_counter()
    for _ in range(iterations):
        table.query(KeyConditionExpression=key_condition_expression)
        print(".", end="")
        sys.stdout.flush()
    print()
    end = time.perf_counter()
    return start, end


if __name__ == "__main__":
    # pylint: disable=not-context-manager
    parser = argparse.ArgumentParser()
    parser.add_argument(
        "endpoint_url",
        nargs="?",
        help="When specified, the DAX cluster endpoint. Otherwise, DAX is not used.",
    )
    args = parser.parse_args()

    test_partition_key = 5
    test_sort_keys = (2, 9)
    test_iterations = 100
    if args.endpoint_url:
        print(f"Querying the table {test_iterations} times, using the DAX client.")
        # Use a with statement so the DAX client closes the cluster after completion.
        with amazondax.AmazonDaxClient.resource(endpoint_url=args.endpoint_url) as dax:
            test_start, test_end = query_test(
                test_partition_key, test_sort_keys, test_iterations, dyn_resource=dax
            )
    else:
        print(f"Querying the table {test_iterations} times, using the Boto3 client.")
        test_start, test_end = query_test(
            test_partition_key, test_sort_keys, test_iterations
        )

    print(
        f"Total time: {test_end - test_start:.4f} sec. Average time: "
        f"{(test_end - test_start)/test_iterations}."
    )

掃描表格以查看用戶端和 Boto3 DAX 用戶端的多次迭代，並報告每個用戶端所花費的時間。

import argparse
import time
import sys
import amazondax
import boto3


def scan_test(iterations, dyn_resource=None):
    """
    Scans the table a specified number of times. The time before the
    first iteration and the time after the last iteration are both captured
    and reported.

    :param iterations: The number of iterations to run.
    :param dyn_resource: Either a Boto3 or DAX resource.
    :return: The start and end times of the test.
    """
    if dyn_resource is None:
        dyn_resource = boto3.resource("dynamodb")

    table = dyn_resource.Table("TryDaxTable")
    start = time.perf_counter()
    for _ in range(iterations):
        table.scan()
        print(".", end="")
        sys.stdout.flush()
    print()
    end = time.perf_counter()
    return start, end


if __name__ == "__main__":
    # pylint: disable=not-context-manager
    parser = argparse.ArgumentParser()
    parser.add_argument(
        "endpoint_url",
        nargs="?",
        help="When specified, the DAX cluster endpoint. Otherwise, DAX is not used.",
    )
    args = parser.parse_args()

    test_iterations = 100
    if args.endpoint_url:
        print(f"Scanning the table {test_iterations} times, using the DAX client.")
        # Use a with statement so the DAX client closes the cluster after completion.
        with amazondax.AmazonDaxClient.resource(endpoint_url=args.endpoint_url) as dax:
            test_start, test_end = scan_test(test_iterations, dyn_resource=dax)
    else:
        print(f"Scanning the table {test_iterations} times, using the Boto3 client.")
        test_start, test_end = scan_test(test_iterations)
    print(
        f"Total time: {test_end - test_start:.4f} sec. Average time: "
        f"{(test_end - test_start)/test_iterations}."
    )

刪除 資料表。

import boto3


def delete_dax_table(dyn_resource=None):
    """
    Deletes the demonstration table.

    :param dyn_resource: Either a Boto3 or DAX resource.
    """
    if dyn_resource is None:
        dyn_resource = boto3.resource("dynamodb")

    table = dyn_resource.Table("TryDaxTable")
    table.delete()

    print(f"Deleting {table.name}...")
    table.wait_until_not_exists()


if __name__ == "__main__":
    delete_dax_table()
    print("Table deleted!")

• 如需詳API細資訊，請參閱〈Python (Boto3)AWS SDK API 參考資料〉中的下列主題。

  • CreateTable

  • DeleteTable

  • GetItem

  • PutItem

  • 查詢

  • 掃描

如需 AWS SDK開發人員指南和程式碼範例的完整清單，請參閱搭配使 DynamoDB AWS SDK。本主題也包含有關入門的資訊以及舊SDK版的詳細資訊。