Overview

Today, I’ll run a hands-on lab on AWS storage services (Amazon S3, Amazon EBS and Amazon EFS).

Hands-on

1. Environment Preparation

1. Launch two private EC2 instances.

Launch two EC2 instances (Amazon Linux 2023) in private subnets-one in each Availability Zone- and attach an IAM role with AmazonSSMManagedInstanceCore.

2. Create Interface Endpoints in two private subnets to allow private EC2 instances to reach Systems Manager (SSM).

Create the following three endpoints to access to the EC2 instances using SSM Session Manager. Replace with your region. com.amazonaws..ssm com.amazonaws..ec2messages com.amazonaws..ssmmessages

Enable Private DNS : ON Security group : create a new one

3. Add rules to the security group attached to the VPC endpoints.

Inbound : HTTPS443 source = the security group attached to the private EC2 Outbound : all traffic (default)

2. EBS hands-on

1. Create an EBS volume

EBS → Volumes → Create volume Type : gp3 Size : 8-10Gib AZ : AZa

2. Attach an additional EBS volume to the EC2 instance

Actions → Attach volume → choose a private the EC2 instance in AZ-a Device name : /dev/sdf

3. Set up the EBS volume

Connect to the EC2 instance by SSM and execute the following commands.

Check block devices.

lsblk

You can see the new EBS volume attached to the instance ("nvme1n1" which hasn’t mounted to anything).

Format the volume (initialize and create a filesystem)

sudo mkfs -t xfs /dev/nvme1n1

Mount (attach it to the filesystem hierarchy) If you record under /data, it’s recorded in the EBS because the EBS volume is mounted to the /data.

sudo mkdir -p /data
sudo mount /dev/nvme1n1 /data

Check that the volume is mounted to the instance. If you see the filesystem mounted on /data, you’ve successfully mounted EBS.

df -h

Check the UUID (= A globally unique identifier assigned to objects in software.).

sudo blkid

Persistence (= Making the OS mount even if it is rebooted) ※Device names may change after reboots, so specify the disk using its UUID.

sudo sh -c 'echo "UUID=xxxx-xxxx  /data  xfs  defaults,nofail  0  2" >> /etc/fstab'

Writing test.

echo "hello ebs" | sudo tee /data/hello.txt
cat /data/hello.txt

4. Create snapshot

EC2 → Volumes → Create snapshot

※You can restore the volume data using snapshot.

3. EFS hands-on

1. Create an EFS file system

VPC : the VPC made in Day1 hands-on Mount targets : two private subnets Security group : create a new SG for EFS with the following settings. Inbound - NFS2049 source = EC2’s SG Outbound - All traffic (default)

2. Mount EFS by EC2

EFS → Attach → Execute the following commands ※Replace with your resource’s id shown on the EFS console.

sudo dnf -y install amazon-efs-utils
sudo mkdir -p /efs
sudo mount -t efs <FILE_SYSTEM_ID>:/ /efs
df -h | grep efs

Do the same steps for both of two EC2 instances to attach EFS for them (EFS can be used for multi-AZ).

3. Verify that both EC2 instances can access the same EFS filesystem.

Execute the following commands by SSM in the EC2 instance in AZa, and create test text file in the folder mounted to the EFS instance.

echo "from ec2-a" | sudo tee /efs/shared.txt
cat /efs/shared.txt

in the EC2 instance in AZb, you can see "shared.txt" file created by the instance in AZa.

cat /efs/shared.txt

4. S3 hands-on

1. Create an S3 bucket

Block Public Access : ON Default encryption : SSE-S3

2. Upload a test file to the bucket.

3. Attach the S3 access permission to the EC2 instance

IAM → Create policy → Set the following JSON to allow EC2 instance to access the S3 bucket made in the previous step ※replace with your S3 bucket name

{
"Version": "2012-10-17",
"Statement": [
{
"Sid": "ListBucket",
"Effect": "Allow",
"Action": "s3:ListBucket",
"Resource": "arn:aws:s3:::<BUCKET_NAME>"
},
{
"Sid": "ObjectRW",
"Effect": "Allow",
"Action": ["s3:GetObject","s3:PutObject","s3:DeleteObject"],
"Resource": "arn:aws:s3:::<BUCKET_NAME>/*"
}
]
}

Attach this policy to the IAM role attached to the private EC2

4. Ensure that the EC2 instance can access the S3 bucket

1. Verify the instance can’t reach the public Internet (because it’s in a private subnet).

curl -I https://aws.amazon.com || true

2. Ensure the instance can reach to the S3 using the S3 endpoint.

※Replace the region name and bucket name with your region

aws s3 ls s3://<BUCKET_NAME> --region <REGION>

You can see the file uploaded on the S3 bucket by console.

3. S3 test operation

Create text file → send it to the S3 bucket →check the S3 folder

echo "hello from private ec2 via vpce" > /tmp/hello.txt

aws s3 cp /tmp/hello.txt s3://$BUCKET/day4/hello.txt --region $REGION
aws s3 ls s3://$BUCKET/day4/ --region $REGION

You can see the file uploaded by the EC2 instance on S3 console.

You can also copy the file from the S3 bucket → check it

aws s3 cp s3://$BUCKET/day4/hello.txt /tmp/hello-from-s3.txt --region $REGION
cat /tmp/hello-from-s3.txt

Tidyng-up

Delete a S3 object → Delete a S3 bucket

Delete EFS File systems (Mount targets are automatically deleted)

Terminate two EC2 instances

Delete EBS volumes

Delete EBS snapshots

※Keep the S3 Gateway endpoint, as it’s no hourly cost.

For test

Key exam points related to today’s services.

Choosing the Right Storage Service

S3 can store objects. Highly durable and requires no scaling

EBS can store block. Low latency, high IOPS. basically for only one AZ and one EC2 instance (can’t be shared with instances)

EFS can store file. can be shared with instances and is multi-AZ.

FSx for windows can store file. SMB, NTFS, ACL, AD integrated (= useful for windows server.)

FSx for Lustre can store file. Integrates with S3. High-speed parallel processing capability

See you soon in Day5!