Cloud Storage Services

When you store files on your laptop, you don't think about whether you'll access them frequently or rarely. The hard drive doesn't care. Cloud storage is different—it makes you predict the future. Every choice about where to put your data is a bet about how you'll use it.

This isn't a limitation. It's an opportunity. Cloud storage lets you optimize for exactly what you need, but only if you understand the tradeoffs you're making.

Three Types of Cloud Storage

Cloud storage comes in three fundamental types, each optimized for different access patterns:

Object storage treats files as independent objects with unique identifiers. You access them via HTTP APIs—request an object by name, get the object back. Amazon S3 pioneered this model; Azure Blob Storage and Google Cloud Storage offer the same approach. Object storage scales to petabytes, costs pennies per gigabyte, and achieves durability so high (99.999999999%) that you're more likely to be struck by lightning than lose data.

The tradeoff: you can't edit part of an object. Want to change one byte in a 10GB file? Upload the entire file again.

Block storage works like a virtual hard drive attached to a virtual machine. AWS EBS, Azure Managed Disks, Google Persistent Disks—they all provide raw blocks that your operating system formats with a filesystem. You get low latency, random access, and the ability to modify files in place.

The tradeoff: block storage costs more, only attaches to one machine at a time, and doesn't scale as gracefully.

File storage provides traditional filesystem semantics—directories, files, permissions—accessible from multiple machines simultaneously. AWS EFS, Azure Files, Google Filestore offer managed NFS or SMB shares.

The tradeoff: highest cost of the three, but sometimes you need shared access to a real filesystem.

The Real Cost Model

Cloud storage pricing has a structure that surprises people who think of storage as "cost per gigabyte."

Storage cost is what you expect—pay per GB stored per month. For standard object storage, this runs $0.02-0.03 per GB. Cheap.

Egress cost is the trap. Uploading data to cloud storage is free. Downloading it costs money—typically $0.08-0.12 per GB. Store a terabyte for a year and pay maybe $300 in storage. Download it once and pay $100 in egress.

This pricing model shapes architectures. It's why you process data in the same region where it's stored. It's why you use CDNs for public content. It's why migrating away from a cloud provider feels expensive—your data is a hostage you have to ransom.

Request costs charge per API call. A few cents per thousand requests. Negligible for most workloads, but matters if you're making millions of small requests.

Retrieval costs apply to archive storage. More on that below.

Storage Tiers: Betting on Access Patterns

Cloud providers offer multiple storage tiers with different cost structures. The fundamental tradeoff:

• Hot/Standard tier: Expensive to store, cheap to access. For data you use constantly.
• Cool/Infrequent Access tier: Cheaper to store (40-50% savings), more expensive to access. For data you touch monthly or less.
• Archive tier: Very cheap to store (70-80% savings), expensive and slow to retrieve. For data you might never need again but can't delete.

Here's where prediction matters. Put frequently-accessed data in archive storage and you'll pay retrieval fees that dwarf the storage savings. Put rarely-accessed data in hot storage and you're paying premium prices for no benefit.

Get it wrong and cloud storage punishes you. Get it right and you can store petabytes for less than the cost of a single hard drive.

Intelligent tiering attempts to solve this by automatically moving objects between tiers based on access patterns. It works well for unpredictable workloads but adds a small per-object fee.

Durability vs. Availability

These sound similar but measure different things.

Durability answers: "Will my data still exist tomorrow?" Cloud storage achieves 99.999999999% durability (eleven nines) by replicating data across multiple devices and facilities. You could store a billion objects and statistically expect to lose one every ten years. This is dramatically more reliable than local storage.

Availability answers: "Can I access my data right now?" Even highly durable storage can be temporarily unavailable. Standard tiers offer 99.9-99.99% availability (a few hours of downtime per year). Lower-cost tiers sacrifice availability for savings.

Replication options affect both:

• Single-datacenter replication: cheapest, vulnerable to facility failures
• Multi-zone replication: survives datacenter failures within a region
• Cross-region replication: survives regional disasters, doubles or triples storage costs

Security That's Built In

Cloud storage security is more sophisticated than what most organizations implement themselves:

Encryption at rest means data is encrypted on the physical storage devices. If someone steals a hard drive from a data center, they get encrypted gibberish.

Encryption in transit means HTTPS/TLS for all data moving over networks. Eavesdroppers can't read your data in flight.

Access control through IAM policies and bucket policies lets you specify exactly who can do what. "This user can read from this bucket but not delete anything." "This application can write to this prefix but not read from others."

Versioning keeps every version of every object. Accidentally overwrite a file? Roll back. Someone deletes data maliciously? Recover it.

Object lock prevents deletion or modification for compliance requirements—useful for regulatory data retention.

Choosing the Right Type

The decision tree is simpler than the options suggest:

Need to attach storage to a VM? Block storage. This is your virtual hard drive.

Multiple machines need shared filesystem access? File storage. Accept the higher cost.

Everything else? Object storage. It's cheaper, more scalable, and more durable.

Within object storage, match tiers to access patterns. If you don't know your patterns, start with intelligent tiering and let the system learn.

The Hybrid Reality

Most organizations don't go all-in on cloud storage. They use cloud strategically:

• Cloud as backup target—offsite protection without managing tapes
• Cloud for cold data—archives that don't justify on-premise infrastructure
• Cloud for burst capacity—overflow when local storage fills
• Gradual migration—move workloads incrementally while maintaining continuity

The key insight: cloud storage isn't replacing traditional storage. It's adding options that didn't exist before.