AWS Lambda function extremely slow to retrieve S3 file

Question

I have a Lambda function that needs to read a file from S3 every time it is executed.
The file is very small, about 200 bytes, the S3 bucket is in the US Standard region, the Lambda function is in the us-east-1 region (so the same region). It takes between 10 and 15 seconds to read the file, how come this is so slow?

Thanks.

EDIT: some code

long start = System.nanoTime();
AmazonS3Client s3Client = new AmazonS3Client();
S3Object propertyFile = null;
try {
    propertyFile = s3Client.getObject(S3_BUCKET_NAME, S3_PROPERTY_FILE);
} catch (Exception e) {...}
try (InputStream in = propertyFile.getObjectContent()) {
    PROPERTIES.load(in);
} catch (Exception e) {...}
LOGGER.debug("S3 access " + (System.nanoTime() - start));

EDIT #2: After Brooks' suggestion I did

AmazonS3Client s3Client = new AmazonS3Client(new InstanceProfileCredentialsProvider());

And I get this error:

Unable to load credentials from Amazon EC2 metadata service

EDIT #3:
The memory allocated to the Lambda function was 256MB, when I allocate 1024MB, it takes 3-4 seconds which is still too slow (it takes about 1-2 seconds when I test locally from my computer).

Answer 1

What I would suggest is to use ProfileCredentialsProvider and to cache S3 client instance between Lambda function executions:

public class MyLambda { // No matter you implement standard AWS SDK interfaces or not

    private final AmazonS3Client s3Client = new AmazonS3Client(new ProfileCredentialsProvider());

    public String sayHello(Request request, Context context) {
        S3Object s3Obj = s3Client.getObject(request.getBucket(), request.getKey());
        return S3Utils.getContent(s3Object); // Some util which retrieves object content
    }
}

The thing is that it takes some time to set up the S3 client which manages connections pool and other resources.