Chapter 44: AWS Storage Comparing
AWS Storage Comparing (or more accurately: how to compare and choose between the different AWS storage services)
There is no single service called “AWS Storage Comparing” — but when people ask this question (especially on YouTube, Reddit India, LinkedIn groups, or in Telugu tech WhatsApp groups), what they really want is:
“Sir, there are so many storage services in AWS — S3, EBS, EFS, FSx, Glacier, Instance Store, FSx for Lustre, Storage Gateway… how do I know which one to pick for my project? Please compare them properly with examples!”
So today we are going to do exactly that — a detailed, honest, practical comparison of the most important AWS storage services that 95% of real-world Hyderabad startups, students, and companies actually use in 2026.
We’ll walk through it like we’re sitting together in a Madhapur café with a big comparison chart on the whiteboard — slow, clear, with everyday analogies, real startup examples, 2026 pricing (ap-south-1 / ap-south-2), and decision rules you can use immediately.
The Big Picture – Why So Many Storage Services Exist?
AWS does not give you “one storage” because different jobs need completely different tools:
- You want to store millions of user-uploaded photos/videos → object storage (S3)
- Your EC2 instance needs a fast boot disk or database volume → block storage (EBS)
- Ten EC2 instances or Lambda functions need to read/write the same folder at the same time → file storage (EFS / FSx)
- You have massive temporary scratch space for ML training or video transcoding → instance store (ephemeral)
- You have 10-year compliance archives you almost never read → cold/archive storage (Glacier / Deep Archive)
- You need Windows file shares or high-performance Lustre for HPC → FSx family
Choosing the wrong one = slow app, high bills, angry users, or compliance problems.
2026 Comparison Table – The Most Important Storage Services
| Service | Type | Persistence | Shared by multiple instances? | Latency / Performance | Durability / Availability | Approx Cost (ap-south-2, 2026) per GB-month | Best Real-World Use Case (Hyderabad 2026) | Never Use It For… |
|---|---|---|---|---|---|---|---|---|
| Amazon S3 | Object | Persistent | Yes (via HTTP API) | Milliseconds (cached via CloudFront: <10 ms) | 11 9s durability, 99.99% avail | Standard: ~₹1.8–2.0 Intelligent-Tiering: ~₹2.0–2.3 Glacier Deep Archive: ~₹0.09 | User photos/videos, static website, backups, data lake, logs | Live databases, root volumes |
| Amazon EBS | Block | Persistent | No (single-attach) / Yes (io2 multi-attach) | Very low (local-like) | 99.999% | gp3: ~₹6.8 io2: ~₹11–15 + IOPS | EC2 boot volumes, RDS/MySQL/PostgreSQL data, high-IOPS apps | Shared read/write by many servers |
| Amazon EFS | File (NFS) | Persistent | Yes (hundreds/thousands) | Milliseconds | 99.999999999% (Standard) | Standard: ~₹21–23 One Zone: ~₹10–12 | Shared wp-content, media library, shared build cache, CMS content | Extremely high IOPS databases |
| Amazon FSx for Windows | File (SMB) | Persistent | Yes | Milliseconds | High | ~₹15–25+ | Legacy Windows apps, Active Directory file shares | Linux-only workloads |
| Amazon FSx for Lustre | File (high-perf) | Persistent | Yes | Very low | High | ~₹20–40+ | ML training, HPC, big-data analytics | General-purpose file sharing |
| EC2 Instance Store | Block (ephemeral) | Temporary | No | Extremely low (local NVMe) | Lower (host-dependent) | Included in instance price | Temporary scratch for ML, video transcoding, Redis cache (replicated) | Persistent data, root volumes |
| S3 Glacier / Deep Archive | Object (cold) | Persistent | Yes | Minutes–hours retrieval | 11 9s | Glacier Flexible: ~₹0.25 Deep Archive: ~₹0.09 | Long-term backups, compliance archives (7–10 years) | Anything you need to access quickly |
4. Real Hyderabad Startup Example – Putting It All Together
Your short-video platform startup in Madhapur (Reels/TikTok style):
Different storage needs & choices:
- User-uploaded raw videos (incoming 1080p files) → S3 Standard bucket “telugu-raw-videos” → Event → Lambda transcodes → output to another S3 bucket
- Transcoded final videos (most watched content) → S3 Intelligent-Tiering + CloudFront → Hyderabad Edge Locations → <10 ms load time
- Old videos >90 days → Lifecycle rule → move to S3 Glacier Instant Retrieval (still fast access if needed)
- Compliance archives (upload logs, KYC videos – 7 years retention) → S3 Glacier Deep Archive + Object Lock (immutable)
- EC2 instances running FFmpeg workers → Root volume: EBS gp3 30 GB → Temporary scratch space for intermediate files: Instance Store (3.8 TB free on c8g.8xlarge)
- Shared thumbnail cache & watermark templates → EFS Standard file system → Mounted on all transcoding containers + designer’s EC2 instances → Everyone sees same files in real time
- Redis cache for trending videos → ElastiCache (uses EBS gp3 under the hood — you don’t manage it)
Monthly storage cost estimate (moderate traffic):
- S3 Standard + Intelligent-Tiering: 3 TB → ~₹5,000–8,000
- Glacier Deep Archive: 5 TB → ~₹450
- EFS Standard: 500 GB → ~₹10,000–12,000
- EBS gp3: 2 TB → ~₹13,000
- Instance Store: ₹0 (included)
- Total storage: ₹30,000–40,000 (vs 2–4× more if using wrong types)
5. Quick Hands-On – Create & Compare Two Storage Types
- S3 (object storage):
- Create bucket “hyd-video-storage-2026”
- Upload short clip.mp4
- Make public (carefully!) → share link
- EBS (block storage):
- Launch t4g.micro
- Create 20 GB gp3 volume → attach → mount → write file
- Stop instance → start again → file still there
- EFS (shared file system):
- Create EFS file system
- Mount on two EC2 instances → write file on one → read on the other
Summary Table – AWS Storage Decision Guide (2026 – India Focus)
| Your Need | Recommended Service(s) | Why This One? (Key Reason) |
|---|---|---|
| Store photos, videos, user uploads, backups | S3 | Unlimited, durable, cheap, CloudFront-ready |
| Fast disk for EC2 boot volume or database | EBS gp3 | Persistent, tunable performance, snapshots |
| Multiple servers/containers need same folder | EFS | Shared NFS, elastic, multi-AZ |
| Windows file shares (SMB) | FSx for Windows | Native AD integration |
| High-performance ML/HPC scratch space | Instance Store + FSx for Lustre | Extreme speed, included/free on certain instances |
| Long-term cheap archive (rare access) | S3 Glacier / Deep Archive | ~₹0.09–0.25/GB-month, compliance-friendly |
Teacher’s final note: AWS Storage is never “one size fits all” — it is “choose the right tool for each job”. In 2026, almost every successful Hyderabad startup uses S3 + EBS + EFS as the core trio, with Glacier/Deep Archive for backups and CloudFront in front of S3 for speed.
Once you understand which service solves which problem, your architecture becomes cheaper, faster, more reliable, and much easier to explain to investors / clients.
Got it? This is the “where should I actually put my data?” master lesson.
Next?
- Deep comparison table with real pricing for 1 TB of each type?
- How to migrate from EBS to EFS or S3 for shared access?
- Or full architecture: S3 + EFS + EBS in one food-delivery-style app?
Tell me — next whiteboard ready! 🚀💾
