starter.wikiSearch⌘K

Data

Key-Value Store

Choose durable key-addressed storage by consistency, persistence, access pattern, geography, limits, and operating model.

Recommendation

Decide whether the values are authoritative before choosing a service.

Redis-compatible state, globally distributed edge reads, and partition-key primary storage have different consistency and durability contracts.13

For: Application and platform teams with a concrete key-addressed state or lookup workload

Main trade-off

Key-addressed systems can simplify and scale declared access patterns, but consistency, durability, data structures, item limits, geography, and portability narrow future use.135

Three key-value storage models

The same key-value label covers data structures, consistency, persistence, geography, and access patterns that are not interchangeable.

  1. Source-of-truth role

    State whether data is authoritative, durable, reconstructable, or merely an acceleration copy.14

  2. Consistency and writes

    Define immediate write visibility, atomic operations, transactions, conditional updates, and conflict expectations.23

  3. Access and limits

    Model keys, values, data structures, item size, indexes, throughput, hot partitions, and scan or query requirements.13

  4. Geography and ownership

    Choose read and write placement, persistence, backup, service coupling, and operational responsibility.23

Storage routes

Choose by the authoritative data contract; send disposable or reconstructable acceleration to Application Caching.

Redis commands and serverless access fit

Evaluate Upstash Redis.

Request-oriented access can provide Redis-compatible data structures without operating Redis infrastructure.

Verify: Validate command compatibility, durability, consistency, regions, request and storage limits, and sustained workload cost.13

Global read distribution matters most

Evaluate Cloudflare Workers KV.

Edge replication can serve read-heavy Workers applications close to users.

Verify: Eventual consistency, write limits, object limits, and Cloudflare coupling constrain frequently mutated authoritative state.123

Access patterns justify a managed primary store

Evaluate Amazon DynamoDB.

Partition- and sort-key design can provide durable managed storage and scale inside an AWS architecture.

Verify: Key design, secondary indexes, item limits, capacity modes, and AWS coupling make unplanned access-pattern changes expensive.1345

Boundary: Application Caching owns reconstructable acceleration. Queues own message delivery and consumer semantics. A broad relational or document system-of-record decision belongs on Database.

Primary-store boundaries

A familiar API name is not evidence of equivalent durability or consistency.

Consistency
Immediate, eventual, and operation-specific guarantees change which values can be authoritative.12
Data structures
Redis commands, opaque edge values, and DynamoDB items support different mutation and query patterns.1
Scale model
Requests, storage, partitions, throughput, regions, and hot keys create different ceilings and cost curves.1
Portability
Redis compatibility, Workers bindings, and AWS data modeling imply different migration effort.12

Official resources

Use the primary documentation to validate product boundaries, operating behavior, limits, and current commercial terms.

Sources

Official documentation supporting the decision routes and their boundaries.

  1. 1
    Upstash Redis getting started

    Upstash · Accessed Official

  2. 2
    Cloudflare Workers KV documentation

    Cloudflare · Accessed Official

  3. 3
    How Workers KV works

    Cloudflare · Accessed Official

  4. 4
    Amazon DynamoDB developer guide

    Amazon Web Services · Accessed Official

  5. 5
    DynamoDB core components

    Amazon Web Services · Accessed Official