RS: new Flex docs #2871
There are no files selected for viewing
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,93 @@ | ||
| --- | ||
| Title: Flex databases | ||
| alwaysopen: false | ||
| categories: | ||
| - docs | ||
| - operate | ||
| - rs | ||
| - rc | ||
| description: Extend Redis databases with flash storage for large-scale, cost-efficient deployments. | ||
| hideListLinks: true | ||
| linktitle: Flex databases | ||
| weight: 38 | ||
| --- | ||
|
|
||
| Flex extends your database capacity by combining RAM and flash (SSD) storage. This tiered architecture keeps frequently accessed (hot) data in RAM for sub-millisecond latency while storing less active (warm) data on flash to reduce costs and increase capacity. Flex frees RAM for hot data, leads to a higher RAM hit rate, and allows bigger datasets per node. | ||
|
|
||
| Flex databases work with your existing Redis applications and the Redis API without modification. | ||
|
|
||
| ## How Flex works | ||
|
|
||
| ### Automatic data tiering | ||
|
|
||
| Flex moves data between RAM and flash based on access patterns: | ||
|
|
||
| - Frequently accessed data stays in high-speed RAM. | ||
| - Less active data moves to cost-efficient flash storage. | ||
| - Data accessed from flash promotes back to RAM automatically. | ||
|
|
||
| Redis uses an [LRU (least recently used)]({{< relref "/develop/reference/eviction#apx-lru" >}}) eviction policy to manage data placement. When memory pressure increases, Flex identifies cold objects, transfers them to flash, and frees RAM for new or frequently accessed keys. | ||
|
|
||
| This process requires no application changes. Your existing Redis commands work across both storage tiers. | ||
|
|
||
| ### Storage engine | ||
|
|
||
| Flex uses [Speedb](https://docs.speedb.io), a high-performance key-value storage engine optimized for flash drives: | ||
|
|
||
| - Redis handles all data operations in memory. | ||
| - Speedb manages the flash storage layer. | ||
|
|
||
| This design delivers predictable latency and throughput as datasets grow beyond RAM limits. | ||
|
|
||
| ## Compatibility | ||
|
|
||
| Flex is compatible with the Redis API and supports all Redis data types, including JSON and probabilistic data structures (Bloom filters, Count-Min Sketch, Top-K). | ||
|
|
||
| The following features are not yet supported with Flex: | ||
|
|
||
| - Redis Search | ||
| - Time series | ||
| - Active-Active | ||
|
|
||
| ## When to use Flex | ||
|
|
||
| Use Flex when you need to: | ||
|
|
||
| - Run Redis at terabyte scale while maintaining high throughput and sub-10 ms latency | ||
| - Power real-time feature stores for machine learning applications | ||
| - Operate large distributed caches with elastic scaling and consistent performance under heavy load | ||
| - Reduce infrastructure costs by combining high-speed RAM with cost-efficient flash storage | ||
|
|
||
| {{<note>}} | ||
| Flex does not replace long-term data persistence. For workloads that require durability and recovery across restarts or failures, use Redis persistence features like [AOF (Append-Only File)]({{< relref "/operate/oss_and_stack/management/persistence#append-only-file" >}}), [RDB snapshots]({{< relref "/operate/oss_and_stack/management/persistence#snapshotting" >}}), or both. For more information, see [Database persistence]({{< relref "/operate/rs/databases/configure/database-persistence" >}}). | ||
|
There was a problem hiding this comment. I suggest adding a dedicated "What Flex isn't" section with the wording similar to the doc: https://docs.google.com/document/d/1PuFejCooJpy_gs0MoM5Pf9MLS46x6gcNUmzYfPjIHik/edit?disco=AAABz_t08H0&tab=t.0
Collaborator
Author
There was a problem hiding this comment. This info is already included in a note in https://redis.io/docs/staging/DOC-6284/operate/rs/flex/#when-to-use-flex:
|
||
| {{</note>}} | ||
|
|
||
| ## Flex and Auto Tiering | ||
|
|
||
| Flex replaces [Auto Tiering]({{< relref "/operate/rs/7.22/databases/auto-tiering" >}}) (formerly known as Redis on Flash). Redis Software selects the implementation based on your Redis version: | ||
|
|
||
| | Redis database version | Flex | Auto Tiering | | ||
| |------------------------|------|--------------| | ||
| | 8.0 and later | <span title="Supported">✅</span> | <span title="Not supported">❌</span> | | ||
| | 7.4 | <span title="Supported">✅</span> | <span title="Supported">✅</span>| | ||
| | 7.2 and earlier | <span title="Not supported">❌</span> | <span title="Supported">✅</span> | | ||
|
|
||
| For Redis Software version 7.22.2-22 or earlier, see [Auto Tiering]({{< relref "/operate/rs/7.22/databases/auto-tiering" >}}). | ||
|
|
||
| ### Differences between Flex and Auto Tiering | ||
|
|
||
| - Key and value offloading | ||
|
There was a problem hiding this comment. Add a short explanation on why it matters: |
||
| - Auto Tiering offloads only values to flash while keys remain in RAM. | ||
| - Flex offloads both keys and values, which increases dataset density per node and reduces RAM consumption. This change frees RAM for hot data, leads to a higher RAM hit rate, and allows bigger datasets per node. | ||
| - RAM population strategy | ||
| - Auto Tiering fills all available RAM before offloading data to flash. This maximizes hot-data performance but can cause non-linear performance changes at high utilization. | ||
| - Flex uses utilization-aware RAM population. When database utilization is below 50%, Flex uses up to 50% of configured RAM for hot data. Above 50% utilization, Flex uses both RAM and flash proportionally, following the configured RAM-to-flash ratio. This provides a stable performance curve, consistent RAM hit-rate, and predictable throughput and latency. | ||
| - Storage engine | ||
| - Auto Tiering uses either RocksDB or Speedb as the storage engine. | ||
| - Flex uses Speedb only. | ||
|
|
||
| ## Next steps | ||
|
|
||
| - [Plan your deployment]({{< relref "/operate/rs/flex/plan" >}}): Review hardware requirements, sizing guidelines, and limitations. | ||
| - [Get started]({{< relref "/operate/rs/flex/get-started" >}}): Configure Flex on your cluster. | ||
| - [Scale your deployment]({{< relref "/operate/rs/flex/scale" >}}): Learn scaling strategies and best practices. | ||
Oops, something went wrong.
Oops, something went wrong.
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
"Flex frees RAM for hot data, higher RAM hit‑rate, bigger datasets per node”
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Added