Friday, June 19, 2026

Cohort 3 | Project CloudIgnite Topics: Cloud Storage Types (Block/Object/File/Hybrid), Amazon EBS, EBS Snapshots, Data Lifecycle Manager (DLM), Instance Store, Amazon EFS, Amazon S3, S3 Storage Classes, S3 Intelligent-Tiering, S3 Versioning, S3 Lifecycle, S3 Object Lock, AWS Storage Gateway, AWS Snow Family, Lab 182 (EBS Volumes & Snapshots), S3 Versioning Lab Duration: ~3 hours

---

Key Takeaways

• The three core cloud storage types are block (EBS), object (S3), and file (EFS/FSx); CloudFront is NOT storage — it's a CDN
• EBS = persistent, single-AZ block storage attached to EC2; every EC2 needs at least one EBS volume for its boot drive
• EBS volume types: SSD (io1/io2 = highest IOPS for databases; gp3/gp2 = general purpose) vs. HDD (st1 = throughput for streaming; sc1 = cheapest cold)
• EBS snapshots are point-in-time, incremental backups stored in S3, can be copied cross-Region, and restored as new volumes; DLM automates creation/retention/deletion (always set retention — 24/day = ~720/month otherwise)
• Instance store = ephemeral (data lost on stop/terminate; survives reboot); use only for cache/buffers/scratch — never for permanent data
• EFS = shared, elastic, NFS, Linux-only file system mounted by many EC2s at once; FSx supports both Windows and Linux
• S3 = object storage with 11 nines durability, strong read-after-write consistency, and buckets/objects/keys; supports event notifications (S3 → Lambda → SNS pattern)
• S3 storage classes trade cost for retrieval: Standard → Intelligent-Tiering → Standard-IA → One Zone-IA → Glacier Instant → Glacier Flexible → Glacier Deep Archive (cheapest, 12–48 hr retrieval)
• S3 Intelligent-Tiering auto-moves objects between tiers based on access pattern (no manual tracking)
• S3 versioning keeps multiple versions; a delete adds a delete marker (must be enabled before the delete to protect the object)
• S3 Object Lock (WORM) supports retention period (N days, cannot delete) and legal hold (indefinite until removed) — used for compliance/governance
• Snow Family = physical devices for offline bulk transfer when bandwidth makes internet transfer impractical (e.g., 50 TB over 100 Mbps ≈ 12.5 MB/s = weeks/months)

---

Table of Contents

1. What Is Cloud Storage?
2. The Three (Plus Hybrid) Storage Types
3. AWS Storage Services Overview
4. Amazon EBS (Elastic Block Store)
5. EBS Snapshots & Data Lifecycle Manager (DLM)
6. Instance Store (Ephemeral Storage)
7. Amazon EFS (Elastic File System)
8. Amazon S3 (Simple Storage Service)
9. S3 Storage Classes & Intelligent-Tiering
10. S3 Features: Versioning, Lifecycle, Object Lock, Security
11. AWS Storage Gateway & Snow Family
12. Lab 182 — EBS Volumes & Snapshots (Console)
13. Lab — EBS Snapshots via CLI + S3 Versioning
14. Glossary
15. Checkpoint Q&A Recap
16. CLF-C02 Exam Relevance Summary
17. Action Items & Housekeeping

---

1. What Is Cloud Storage?

Cloud storage = a service that stores your data on the internet through a cloud provider that manages and operates the underlying storage. Data lives in the provider's data centers and is accessed over an internet connection (not on your local disk).

CLF-C02 Relevant (Medium): Understanding that cloud storage is a managed, internet-accessible service is foundational to Domain 3 (Cloud Technology).

---

2. The Three (Plus Hybrid) Storage Types

• Block (EBS): suitable for operating systems, applications, and databases. Every EC2 instance needs at least one EBS volume (its "C: drive").
• Object (S3): stores any file as an object. Not the same as "file storage."
• File (EFS / FSx): a true shared file system.
• CDN (CloudFront) is NOT storage — it's a content delivery service (a common point of confusion).

Visual: The three storage types (plus hybrid)

The three core storage types and the AWS service each maps to.

CLF-C02 Relevant (High): Knowing block vs. object vs. file storage and mapping them to EBS / S3 / EFS is a very common exam question.

---

3. AWS Storage Services Overview

Important: Frequently-asked distinction: EFS is Linux-only, while FSx supports both Windows and Linux.

---

4. Amazon EBS (Elastic Block Store)

• Persistent block storage attached to EC2 (survives stop/start, unlike instance store).
• EBS is a single-AZ (zonal) service — NOT regional. A volume can only attach to an EC2 instance in the same Availability Zone.
• Can scale size up/down within minutes (e.g., 8 GB → 20 GB).
• Supports encryption at rest and snapshots.
• Common use cases: boot volumes (OS), data storage with a file system, and database hosting (RDS uses EBS under the hood).

Visual: EBS is single-AZ (zonal)

An EBS volume can attach only to an EC2 instance in the same Availability Zone.

EBS volume types

• Cost decreases left-to-right: io2 (most expensive) → gp → st1 → sc1 (cheapest).
• IOPS vs. Throughput: IOPS = number of read/write operations per second (matters for databases); Throughput = amount of data per second (matters for streaming large files).

Visual: Choosing an EBS volume type

SSD for IOPS/boot, HDD for throughput or cheap bulk; cost falls io2 → gp → st1 → sc1.

CLF-C02 Relevant (High): "EBS is block storage, single-AZ, persistent" and "SSD = IOPS/databases, HDD = throughput/streaming & cheap" are exam favorites.

---

5. EBS Snapshots & Data Lifecycle Manager (DLM)

Snapshots

• A snapshot = a point-in-time backup of an EBS volume, stored in S3 (managed by AWS).
• Incremental: only changed blocks are saved after the first snapshot — saves space/cost.
• You can copy a snapshot to another Region, then create a new volume from it there — useful for deploying the same data/EC2 in a different Region.
• Restore = create a new volume from a snapshot (like restoring from backup).

Data Lifecycle Manager (DLM)

• A service to automate creation, retention, and deletion of EBS snapshots (e.g., one every hour).
• Snapshots cost money — automating without deletion can produce 24/day = ~720/month and a surprise bill. Always set retention (e.g., keep only the last 2–3).
• Uses tags to identify volumes and a lifecycle policy to define retention.
• Requires an IAM role — specifically the AWS Data Lifecycle Manager default role (aws dlm create-default-role).

Visual: EBS snapshot lifecycle

Snapshots are incremental, S3-backed backups you can copy cross-Region or restore into a new volume; DLM automates creation and retention.

CLF-C02 Relevant (High): Snapshots as S3-backed, incremental, cross-Region-copyable backups — plus DLM automation and the cost/retention message — align with reliability and cost-optimization themes.

---

6. Instance Store (Ephemeral Storage)

• Temporary, non-persistent block storage physically attached to the host of the EC2 instance → very fast, low latency.
• Bound to the instance lifecycle: stop or terminate → data is lost. A reboot is OK (data persists).
• Cannot be detached/managed independently of the instance (unlike EBS).
• Availability and size depend on the instance type (not available on all types).
• Must be mounted before use (auto or manual on Linux).
• Use cases: caching, buffers, scratch/temporary data, and data replicated across a fleet (e.g., a load-balanced pool of web servers). Never for permanent data, OS, or important files.

Visual: Instance store lifecycle

Instance-store data survives a reboot but is lost on stop or terminate — use it only for cache, buffers, and scratch data.

CLF-C02 Relevant (Medium): "Instance store = ephemeral, lost on stop/terminate, good for cache/temp" vs. "EBS = persistent" is a known comparison.

---

7. Amazon EFS (Elastic File System)

• A scalable, fully managed, elastic network file system (NFS) — petabyte-scale, low latency.
• Shared: one EFS file system can be mounted by many EC2 instances at once (block storage like EBS is not shared). Acts like a network drive ("a big Samba share").
• Linux only (uses the NFS protocol). Can also connect to on-premises servers.
• Auto-scales: grows and shrinks automatically as files are added/removed — no capacity management.

EFS storage classes & performance

• EFS Standard → frequent access.
• EFS Standard-Infrequent Access (IA) → cheaper for rarely-accessed files.
• EFS One Zone → cheapest; single AZ; use only for reproducible data (data lost if the zone fails).
• Performance modes: General Purpose, Max I/O.
• Throughput modes: Elastic, Bursting (baseline + burst), Provisioned (fixed).
• Use cases: home directories, enterprise app file systems, app testing/deployment, database backups, content management, media workflows, big data analytics.
• To use: create the EFS file system → create a mount target → mount it on EC2 → use like a network drive.

Visual: EFS is a shared file system

One EFS file system can be mounted by many EC2 instances (and on-prem servers) at once — unlike EBS, which attaches to a single instance.

CLF-C02 Relevant (Medium–High): "EFS = shared, elastic, NFS, Linux-only file system" — and that multiple EC2s can mount it concurrently — is commonly tested.

---

8. Amazon S3 (Simple Storage Service)

• Object storage service — secure, durable, highly available; effectively unlimited capacity.
• Famous for eleven 9s (99.999999999%) of durability.
• Store/retrieve any amount of data, anytime, from anywhere.
• Strong read-after-write consistency.

Core concepts

• Bucket = a container for objects; created in a specific Region.
• Object = the stored item; identified by a key that must be unique within the bucket.
• Supports event notifications that can trigger SQS, SNS, or Lambda (e.g., the S3 → Lambda → SNS pattern from a previous lecture).
• CLI commands: aws s3 ls, cp (upload/download), sync (with/without --delete), rm, mb (make bucket).

CLF-C02 Relevant (High): S3 as durable (11 nines) object storage, buckets/objects/keys, Region scope, and event integration are core exam material.

---

9. S3 Storage Classes & Intelligent-Tiering

Intelligent-Tiering (most-emphasized topic)

• Monitors each object's access pattern and automatically moves it: frequently accessed → Standard; less accessed → Standard-IA; rarely accessed → Glacier. No manual tracking.

Cost model reminder

• IA / Glacier classes charge less for storage per GB but more for access/retrieval. Frequent access to an IA/Glacier object can cost more than Standard.
• Archive tip: don't archive many tiny files — zip them into one file to reduce per-access cost.

Visual: S3 storage classes — cost vs. retrieval

Storage cost drops left → right while retrieval gets slower; Intelligent-Tiering auto-moves objects so you don't have to.

CLF-C02 Relevant (High): The instructor flagged storage classes — especially Intelligent-Tiering, One Zone-IA, and the Glacier tiers (including Deep Archive's 12–48 hr retrieval) — as among the most frequently asked exam topics.

---

10. S3 Features: Versioning, Lifecycle, Object Lock, Security

Versioning

• Keeps multiple versions of an object. Deleting an object adds a delete marker instead of erasing it — you can restore previous versions.
• Not enabled by default; states are Disabled / Enabled / Suspended.
• Versioning only protects objects deleted after it's enabled (lesson learned in the lab).

Lifecycle management

• Automatically transition objects between tiers or delete them on a schedule (e.g., after 30 days → IA, after 60 days → Glacier, then delete). Can be automatic (Intelligent-Tiering) or a manual lifecycle policy.

Object Lock (WORM — compliance/governance)

• Retention period: lock a file so it cannot be deleted for N days (e.g., 30 days).
• Legal hold: lock a file indefinitely until the hold is removed.
• Important for compliance and governance.

Security & access

• Objects are private by default; you can grant public access or block public access (account/bucket level).
• Pre-signed URLs: grant temporary, time-limited access to a private object.
• CORS (cross-origin resource sharing): needed when S3 hosts a static website.

Visual: S3 versioning & restore

With versioning on, S3 keeps every version and a delete only adds a delete marker — so you can restore a prior version (it must be enabled before the delete to protect that object).

CLF-C02 Relevant (High): Versioning, lifecycle policies, Object Lock (compliance), block-public-access, and pre-signed URLs are all exam-relevant S3 security/management features.

---

11. AWS Storage Gateway & Snow Family

AWS Storage Gateway (Hybrid)

• Connects an on-premises data center with AWS storage so both are used together (not a one-time migration/replication). Use cases: disaster recovery, tiering, migration.

AWS Snow Family (Physical Data Transfer)

• Physical devices AWS ships to you; you copy data onto them and ship them back to an AWS data center.
• Why not just use the internet? Large transfers take too long. Example: 50 TB over a 100 Mbps link — since 100 Mbps ≈ 12.5 MB/s, this would take weeks/months. Snow devices are faster for hundreds of TB to PB.
• Device size scales with data volume (up to truck-scale for petabytes).

Visual: Snow Family vs. internet transfer

When bandwidth makes an internet transfer take weeks (e.g., 50 TB over 100 Mbps), ship the data physically with the Snow Family.

CLF-C02 Relevant (Medium): Recognize Storage Gateway (hybrid) and Snow Family (offline bulk migration when bandwidth is insufficient) as the right tools for those scenarios.

---

12. Lab 182 — EBS Volumes & Snapshots (Console)

Goal: Create an EBS volume, attach/mount it to EC2, write data, snapshot it, delete the data, and restore from the snapshot — all via the Management Console + Linux CLI on the instance.

Key steps:

1. Note the running EC2 instance's Availability Zone (EC2 → Networking) — e.g., us-west-2a.
2. Create an EBS volume (gp2, 1 GB) in the same AZ as the instance. (Different AZ → cannot attach.)
3. Attach the volume (Actions → Attach volume) as device /dev/sdb.
4. Connect via EC2 Instance Connect; run df to view disks (EBS shows as /dev/nvme...).
5. Create a file system: mkfs -t ext3 /dev/sdb (a raw volume isn't usable until formatted — like formatting a USB drive with FAT32/NTFS; Linux uses ext3/ext4).
6. Mount it: create /mnt/datastore, then sudo mount /dev/sdb /mnt/datastore. Confirm with df.
7. Create files in the mounted folder (data now lives on the new volume).
8. Create a snapshot (Volumes → Actions → Create snapshot).
9. Delete the files (rm *.txt), then restore: create a new volume from the snapshot (same AZ), attach as /dev/sdc, mount to /mnt/datastore2.
10. The restored volume keeps the same file system (no reformat needed) and the original files are back.

Visual: Lab 182 — EBS volume & snapshot workflow

Create, attach, format, mount, snapshot, delete, then restore a new volume from the snapshot with the file system and data intact.

Instructor: "This lab was a lot easier than the others." The next lab does the same thing via AWS CLI.

---

13. Lab — EBS Snapshots via CLI + S3 Versioning

Part A — EBS snapshots via AWS CLI:

1. Attach an instance profile / IAM role ("S3 bucket access") to the EC2 instances (Actions → Security → Modify IAM role) so they can reach S3.
2. Connect to the command host via EC2 Instance Connect.
3. describe-instances / describe-volumes to find the volume ID and instance ID.
4. Because the target EBS is a root volume, stop the instance before snapshotting it (stop-instances), confirm it's stopped, take the snapshot, then start it again.
5. Scheduled snapshots without stopping: use a cron job to run a snapshot command every minute — demonstrating snapshots piling up fast.
6. Run a Python (boto3) script (snapshotter.py) that lists/sorts snapshots by start time and keeps only the last 2, deleting older ones — reinforcing the cost/retention lesson. (Requires installing the boto3 module; run with python3.8.)

Part B — S3 versioning challenge:

1. Create an S3 bucket (globally unique name).
2. Download and unzip sample files (file1, file2, file3) on the processor instance.
3. Enable versioning on the bucket before uploading.
4. aws s3 sync the files up to the bucket; verify with aws s3 ls.
5. Delete a file locally and sync --delete → S3 shows a delete marker (object still exists as a prior version).
6. Restore the deleted object using its version ID (get-object with --version-id), then sync (without --delete) to put it back.

Warning: A student couldn't restore file1 because versioning was enabled after the file was first deleted — only file2 (deleted after versioning was on) could be restored. Enable versioning first.

CLF-C02 Relevant (Medium): The labs reinforce EBS snapshots, automation/cost control, IAM roles for service access, the AWS SDK (boto3), and S3 versioning + restore — all exam-relevant concepts (the CLI mechanics themselves are not tested).

---

14. Glossary

---

15. Checkpoint Q&A Recap

• Low-cost, long-term archive/backup storage? → S3 Glacier.
• Which service is block storage? → EBS.
• Archived data that needs immediate access? → S3 Glacier Instant Retrieval.
• Volume type for additional storage on a virtual desktop? → gp3 (General Purpose SSD) (or HDD if not a boot volume).
• Prevent latency after restoring a volume from a snapshot / automate backups? → use DLM for automatic creation, retention, and deletion.
• IAM role needed before using DLM? → the AWS Data Lifecycle Manager default role.
• Three components of cloud storage? → client device, internet, data center.
• Logical network segment in a VPC within a single AZ? → Subnet (carryover review).
• Which file system protocol does EFS use? → NFS.
• Prevent objects from being deleted? → Object Lock.
• S3 versioning states? → Disabled, Enabled, Suspended.

---

16. CLF-C02 Exam Relevance Summary

---

17. Action Items & Housekeeping

• Submit both labs before leaving (instructor repeated this).
• Optional (0-point) lessons: complete the interview preparation guideline and the cloud-storage lesson (recommended, ~10 min each).
• Complete the S3 versioning challenge on your own if not finished in class.
• Upcoming topics (Monday): S3 Glacier, Storage Gateway, more work with Amazon S3, an S3 challenge, and the Snow Family.
• Schedule note: ~6 class days remain; a lab session Saturday 2–6 PM; some content slips to July; instructor will attend the graduation ceremony.

---

Notes compiled from the June 19, 2026 session transcript for post-lecture review — AWS re/Start, Cohort 3: Project CloudIgnite.