This document will guide you in automating the collection of RDS/Aurora Cloudwatch metrics for assessing your database workload(s). The collected metrics can be used to provide placement guidance for right sizing purposes. You can run this automation either from an EC2 instance or AWS CloudShell environment.

Cloudshell > Actions > Upload file

shell> chmod a+x tessell_fetch_rds_aurora_metrics.sh

shell> bash tessell_fetch_rds_aurora_metrics.sh -h

The following are required to run this automation:

To install the dependent packages and automation, follow these steps:

getopt

AWS CLI

Installation instructions can be found here.

zip

Deploy Code

Copy the attached file “tessell_fetch_rds_aurora_metrics.sh” to a location where you plan to execute the automation and make the file executable.

shell> chmod a+x tessell_fetch_rds_aurora_metrics.sh

Verify Installation

To validate the installation of dependent packages and the automation code, follow these steps:

shell> which getopt

shell> which aws

shell> which zip

shell> bash tessell_fetch_rds_aurora_metrics.sh -h

Setup AWS Credentials

Configure aws credentials in the host environment where you run this code.

You can get the aws programmatic access from AWS Console and for more details, please refer to the link.

shell> aws configure

AWS Access Key ID : [****************]

AWS Secret Access Key : [****************]

Default region name : us-east-1

Default output format : None

Create a text file with the list of RDS/Aurora Database identifier name(s) separated by a new line.

You can get the DB identifier name from the AWS RDS console.

You can specify n number of db instances in the text file but ensure that they run in the same region. If the instances are in different regions, you need to gather the information separately.

Example: If you have one RDS and one Aurora cluster with 1 writer and 4 reader nodes, then the file should look like the following:.

cat /tmp/db_list.txt

my-rds-service

my-cluster-writer

my-cluster-reader1

my-cluster-reader2

my-cluster-reader3

my-cluster-reader4

Run the automation with the number of days to be analyzed along with Cloudwatch metric sampling interval. This will take a few minutes depending upon the number of db instances to be analyzed.

Suppose if you want to gather metrics for Aurora cluster for last 7 days with sampling interval 86400 seconds, you can run as follows:

shell> time bash tessell_fetch_rds_aurora_metrics.sh --type aurora --days 7 --period 86400 --file /tmp/db_list.txt

The generated logs and metrics can be found in /tmp/get_metrics* folder. You can import the generated CSV file into an Excel sheet for better visibility and for performing db workload analysis with ease.

Open Google Sheets > File > Import > Upload > Drop csv file

tessell_fetch_rds_aurora_metrics.sh --type <type> --file <input-file> --period <period> --days <number of days>

where,

--help | -h : Display this help.

--type | -t : Please provide db type either RDS or AURORA to collect metrics[Mandatory].

--file | -f : Please provide a file containing a db identifier list separated by new lines[Mandatory].

--days | -d : Please provide a number of days to determine the first data point and last data point for metrics.

--start_date | -s : Please provide a start date to determine the first data point in the format 'YYYY-MM-DDTHH:MM:SS'.

--end_date | -e : Please provide an end date to determine the last data point in the format 'YYYY-MM-DDTHH:MM:SS'.

--period | -p : The granularity, in seconds, of the returned data points. Default value is 86400 seconds.

Example:

tessell_fetch_rds_aurora_metrics.sh --type <type> --file /path/to/db_list.txt --period <period> --days <number of days>

tessell_fetch_rds_aurora_metrics.sh --type <type> --file /path/to/db_list.txt --period <period> --start_date '2020-12-10T00:00:00' --end_date '2020-12-10T23:59:59'