Chapter 7 - Elastic Block Storage and Elastic File System

Question 1

EFS

Answer 1

1. EFS stands for Amazon elastic file system
2. Basically, it’s a managed NFS, or network file system, that can be mounted on many different EC2 instances at once.
3. So EFS works with EC2 instances that are in multiple availability zones.
4. So you can set up EFS and have centralized storage and attach it to multiple EC2 instances.
5. So it’s basically just shared storage.
6. It’s highly available
7. it’s scalable,
8. it’s expensive.
9. So this is a great way of having web server farms.
10. It’s a great way for things like content management systems, or shared database access, etc.
11. So it’s a shared storage across multiple EC2 instances.
12. It uses the NFSv4 protocol,
13. It’s compatible with Linux-based AMIs,
14. Windows is not supported at this time.
15. It uses encryption at rest using key management service.
16. It’s a file system that scales automatically, and you don’t have to worry about capacity plannings.
17. EFS has amazing performance capabilities, however, you can have thousands of concurrent connections.
18. EFS can literally support thousands of concurrent connections or thousands of EC2 instances.
19. You can get 10 Gbps throughput, so it’s really super fast in terms of throughput,
20. It scales to the petabyte.

Question 2

EFS - Exam Tips

Answer 2

1. So EFS supports the network file system version 4, NFSv4 protocol.
2. You only pay for the storage that you use, so you don’t need to worry about pre-provisioning like we just saw then.
3. You click it, and it’s there almost instantly.
4. It can scale up to petabytes
5. It can support thousands of concurrent NFS connections at once.
6. And data is stored across multiple availability zones within a region.
7. And you get read-after-write consistency.

If you have a scenario-based question around highly scalable shared storage using NFS, I want you to think immediately of EFS.

Question 3

FSx

Answer 3

You’ll be given different scenarios and asked to choose whether you should use EFS, FSx for Windows, or FSx for Lustre.

1. Use EFS where you need distributed, highly resilient storage for Linux instances and Linux-based applications.
2. Amazon FSx for Windows is where you need centralized storage for Windows-based applications, Essentially anything that’s Microsoft,
3. It needs centralized storage, just straightaway, think of Amazon FSx for Windows.
4. Amazon FSx for Lustre is where you need high-speed, high-capacity distributed storage.

• And this will be for applications that do high performance computing, financial modeling, etc, etc.
• And remember that FSx for Lustre can store data directly on S3.

Question 4

EBS

Answer 4

SSD volumes

• They’re highly available and scalable storage volumes that you can attach to an EC2 instance.

gp2

• This is general purpose SSD.
• This is suitable for boot disks and general applications.
• Gives you up to 16000 IOPS per volume,
• It gives you up to 3 nines durability.

gp3

• This is suitable for high performance applications
• And it gives you a predictable 3000 IOPS baseline performance
• And 125 MiB per second, regardless of your volume size.
• This is 99.9% durability, so 3 nines durability.

io1

• If you need a faster SSD volumes
• This is provisioned IOPS SSD.
• This is suitable for online transaction processing and latency-sensitive applications.
• You get up to 50 IOPS per GiB
• and up to 64000 IOPS per volume.
• It’s high performance but it’s also the most expensive EBS volume that you can get.
• And you get 3 nines durability.

iO2

• Is the the new provisioned IOPS SSD.
• Is suitable for online transaction processing and latency-sensitive applications.
• You get 500 IOPS per GiB
• and about 64000 IOPS or up to 64000 IOPS per volume.
• And with io2, you actually get 5 nines durability.
• So it has the best durability out of any EBS volumes.
• It’s the latest generation provisioned IOPS volume.

magnetic storage

• So we had throughput optimized hard disk drive.
• This is suitable for big data, data warehouses, extract, transform, and load.
• The maximum throughput is 500 MB per second,
• It cannot be a boot volume
• You do get up to 3 nines durability.

sc1

• This has a max throughput of 250 MB per second, per volume.
• It gives you less frequently accessed data.
• It cannot be a boot volume
• It is the lowest cost
• It gives you 3 nines durability.
• So just remember that volumes exist on EBS,
• Whereas snapshots exist on S3.
• Snapshots are basically like what you would do with a camera.
• They’re a point-in-time photograph of a volume
• They’re incremental in nature.
• The first snap that you will take is going to take some time to create
• For consistent snapshots you should really stop your instance and detach the volume. That way everything is then saved to disk.
• You can share your snaps between AWS accounts,
• Between regions
• but first you need to copy that snapshot to the target region.
• And you can resize EBS volumes on the fly, as well as changing the volume type. So you can go from gp2 to gp3, for example.
• So instant store volumes are sometimes called ephemeral storage.
• And you can reboot both EBS instances and instance store volumes and you will not lose your data.
• However, instance store volumes cannot be stopped.
• So if the underlying host fails, you are going to lose your data with instance store volumes.
• By default, both root volumes will be deleted on termination.
• However, with the EBS volumes, you can actually tell AWS to keep the root device volume.
• EBS backed instances can be stopped.
• So you’re not going to lose any data on your instance if it is stopped.