Dynamodb阅读性能差

Question

These days, I have tested Dynamodb performance and I have been disappointed at the performance.

I used query command on just partition-key without any filter condition. It means it just simply reads an index and return.

dynamodb.query(
    TableName='test-table',
    ReturnConsumedCapacity='TOTAL',
    KeyConditionExpression='tid = :tid',
    Limit=5000,
    ExpressionAttributeValues={
        ':tid': {'N': '9999999'},
    }
)

---------- Part of Response ----------
{
    .
    .
    .
    "Count": 5000,
    "ScannedCount": 5000,
    "LastEvaluatedKey": {
        "tid": {
            "N": "9999999"
        },
        "uid": {
            "N": "985377"
        }
    },
    "ConsumedCapacity": {
        "TableName": "sam-push-test",
        "CapacityUnits": 110.5
    }
}

It takes about 0.5 sec. It is tested in AWS Lambda(1024 memory) located in the same region as Dynamodb.

I am curious that poor performance is a normal situation or there should be something I have missed.

Apparently, AWS is saying about Dynamodb delivers single-digit millisecond performance.

Amazon DynamoDB is a key-value and document database that delivers single-digit millisecond performance at any scale.

In addition, query-latency metric in Dynamodb console shows 100-150ms on my test.

How is it measured?

Answer 1

You should expect single digit millisecond performance when reading or writing a single item. But when querying a page of 5000 items (as it seems to be in your case) it's normal to see ~100ms.

Of course, you should also make sure that the client is not too far away from the DynamoDB endpoint. A round trip around the world can easily add more than 100ms.