Block, file and hybrid storage, and AWS Backup: EBS, instance store, EFS, FSx and Storage Gateway

Why this matters

Lesson 18 covered Amazon S3, where data is stored as objects and reached over the internet through an API. Most business software was not written that way. A database expects a disk it can write to directly. A team of Windows users expects a shared drive with folders and permissions. A backup application expects a tape library. AWS offers a different storage service for each of these expectations, and choosing the wrong one is a common, expensive mistake: data that vanishes when a server stops, a file share that cannot be reached from a second server, or a backup that nobody actually takes.

A manager or salesperson needs this lesson to follow a migration conversation. When someone says "we lift the file server as it is" or "keep the tapes but put them in the cloud", there is a specific AWS service behind each phrase. A junior engineer needs it because the line between temporary and persistent storage decides whether a restart is harmless or a disaster.

The lesson works through four families: block storage attached to a single server, shared file systems, hybrid storage that keeps a local cache in the customer's own building, and the central backup service that protects all of them.

The lesson

Block storage: Amazon EBS and instance store

Block storage splits data into fixed-size chunks called blocks, and the operating system decides how to arrange them. To a server, a block device looks exactly like a hard drive: it is formatted with a file system, the operating system boots from it, and a database writes to it at low latency. Block storage is attached to one server at a time, which is the main way it differs from the file services below.

Amazon Elastic Block Store (Amazon EBS) provides block storage volumes for Amazon EC2 instances. Three facts about EBS answer most questions about it.

• It is persistent. An EBS volume is reached over the network, not built into the physical host, so its data survives when the instance is stopped and started. A volume can even be detached from one instance and attached to another. This is why EBS is the normal choice for a boot volume and for a database running on EC2.
• It lives in one Availability Zone. A volume is created in a single zone and attaches only to instances in that same zone. AWS replicates the volume inside that zone to protect against hardware failure, but that replication is not a backup and does not protect against an entire zone failing or against someone deleting data.
• It is backed up with snapshots. A snapshot is a point-in-time copy of a volume, held by AWS in Amazon S3 behind the scenes. Snapshots are incremental: after the first, each one stores only the blocks that have changed. A new volume can be created from a snapshot, including in another Availability Zone, and snapshots can be copied to other Regions. That is how EBS data moves between zones.

A volume has a size the customer provisions, and the customer chooses a volume type: SSD-backed types suit transactional databases and boot volumes, while HDD-backed types suit large, sequential workloads. A feature called EBS Multi-Attach lets certain volume types attach to several instances, but only within one zone, so it is not a substitute for a shared file system.

An instance store is also block storage, but it sits on disks physically attached to the host computer that runs the instance. That closeness makes it very fast, and its cost is included in the instance price. The trade-off is that it is temporary: data in an instance store is lost when the instance is stopped, hibernated or terminated, or when the underlying disk fails. It survives a simple reboot. Not every instance type has one. Instance store suits buffers, caches, scratch data and copies of data that are already replicated somewhere else.

Exam framing: words such as "persist", "survive a stop" or "boot volume" point to EBS. Words such as "temporary", "scratch" or "acceptable to lose" point to instance store.

Shared file systems: Amazon EFS and Amazon FSx

File storage organises data as files in folders, and many servers can open the same files at the same time over a network protocol. It is the cloud version of the shared drive found in most offices.

Amazon Elastic File System (Amazon EFS) is a fully managed file system that uses NFS (Network File System), the protocol Linux servers use for shared files. Its defining features are these:

• Many EC2 instances, in multiple Availability Zones within a Region, can mount the same file system at the same time. Containers and Lambda functions can use it too, and on-premises servers can reach it over a private connection.
• It is elastic: it grows and shrinks automatically as files are added and removed, so there is no size to provision and the customer pays for the storage actually used.
• AWS runs the servers, the replication and the scaling.

Amazon FSx is a family of fully managed file systems, each built on a well-known commercial or open-source file system. It exists for workloads that need a specific file system that EFS does not provide.

• Amazon FSx for Windows File Server gives Windows applications native file shares over SMB (Server Message Block, the Windows file-sharing protocol), integrated with Microsoft Active Directory for users and permissions.
• Amazon FSx for Lustre is a high-performance file system for compute-heavy work such as high performance computing and machine learning, and it can link to data in Amazon S3.
• Amazon FSx for NetApp ONTAP gives customers who already run NetApp storage the same ONTAP data management features as a managed service.
• Amazon FSx for OpenZFS does the same for the open-source OpenZFS file system.

Exam framing: "Linux", "NFS", "shared across instances in several zones" or "grows automatically" points to EFS. "Windows", "SMB", "Active Directory", "Lustre" or "NetApp ONTAP" points to FSx.

Hybrid storage with a local cache: AWS Storage Gateway

Some data cannot simply move. Applications in the customer's own data centre (on premises) may need to keep using local file shares, disks or tape drives while the business wants the storage itself, and its durability, in AWS. A cached file system solves this: the full data set is stored in AWS, while a local copy of the most recently used data is kept close to the applications, so frequent reads stay fast.

AWS Storage Gateway is the hybrid storage service that does this. It runs on premises, usually as a virtual machine, and presents familiar storage interfaces to local applications while storing the data in AWS. It comes in several types:

• Amazon S3 File Gateway presents file shares over NFS or SMB. Files written to the share are stored as objects in Amazon S3, and recently used files are cached locally for low-latency access.
• Volume Gateway presents block storage volumes to on-premises servers using the iSCSI protocol. In cached mode the primary data lives in AWS with frequently used data held locally; in stored mode the full data set stays local and is copied to AWS asynchronously as EBS snapshots.
• Tape Gateway presents a virtual tape library to existing backup software. The company keeps its backup software and processes, but the virtual tapes are stored in Amazon S3 and can be archived to the S3 Glacier storage classes. It is the answer when a company wants to stop handling physical tapes.

Two near-misses catch candidates. AWS Direct Connect provides a private network connection to AWS; it can carry Storage Gateway traffic, but it stores nothing and presents no file shares. AWS Transit Gateway, despite its name, connects VPCs and on-premises networks and has nothing to do with storage.

Understanding use cases for AWS Backup

Every service above can be protected in its own way, and a large organisation might have hundreds of EBS volumes, databases and file systems, each with its own settings. AWS Backup is a fully managed service that centralises and automates backups across many AWS services from one place. It supports, among others, Amazon EBS volumes, Amazon EC2 instances, Amazon RDS and Amazon Aurora databases, Amazon DynamoDB tables, Amazon EFS and Amazon FSx file systems, Amazon S3 and Storage Gateway volumes.

Its main ideas:

• A backup plan is a policy that says how often to back up, how long to keep each backup, and when to move it to colder, cheaper storage. Resources are assigned to a plan, often by tag, so a new database tagged correctly is protected automatically.
• Backups are stored as recovery points in a backup vault. AWS Backup Vault Lock can make a vault's backups impossible to delete or change for a set retention period, which helps against ransomware and meets write-once retention rules.
• Backups can be copied to another Region or another account, and with AWS Organizations, backup policies can be applied across many accounts.
• Reporting and audit features show whether resources are actually protected, which supports compliance.

Typical use cases are meeting a regulator's retention requirement, recovering from accidental deletion or ransomware, and replacing a collection of scripts with one consistent policy.

Do not confuse it with AWS Elastic Disaster Recovery, which continuously replicates whole servers so they can be launched quickly in AWS after an outage. That is recovery of running systems, not scheduled, retained backups of databases and file systems.

Shared responsibility applies here, as lesson 6 explained. AWS keeps the EBS service running and replicates volumes within their zone. Protecting the customer's data, by taking snapshots or using AWS Backup plans, is the customer's job. No support plan takes backups on a customer's behalf.

What the exam is really asking

Most scenarios in this area are a matching exercise between a need and a storage shape:

• One server, needs a disk that persists: EBS.
• One server, needs the fastest temporary disk: instance store.
• Many Linux servers sharing files across zones: EFS.
• Windows shares with Active Directory, or Lustre, ONTAP or OpenZFS: FSx.
• On-premises applications using AWS storage through a local cache, or virtual tapes: Storage Gateway.
• One policy to back up many services: AWS Backup.

Pricing for these services is covered in lesson 23; check the current figures on each service's pricing page or in the AWS Pricing Calculator rather than memorising them.