Skip to content

DOC-7142 Update RDI when-to-use limits for the Flink default; reunify Cloud and on-prem lists - #4176

Merged
andy-stark-redis merged 2 commits into
mainfrom
DOC-7142-update-rdi-documentation-for-flink-default-in-version-2-0-0
Oct 2, 2026
Merged

andy-stark-redis merged 2 commits into
mainfrom
DOC-7142-update-rdi-documentation-for-flink-default-in-version-2-0-0

Conversation

@andy-stark-redis

@andy-stark-redis andy-stark-redis commented Oct 1, 2026 •

Copy link
Copy Markdown
Contributor

DOC-7142

The Flink processor is the default from RDI 2.0.0. This PR updates the "When to use RDI" limits to match, and puts the on-prem and Redis Cloud pages back on the same embed.

Changes

  • Limits: the on-prem list now uses the same Flink numbers as the Cloud page: 20K changes/s, 60K records/s for full sync, 20K for CDC, and 200GB. The decision tree uses the same thresholds now.

  • Note: reworded as Mirko suggested. The limits apply to the Flink processor, which arrived in 1.18.0 and is the default from 2.0.0. The classic processor supports about half of each limit.

  • One shared embed for both pages: with the numbers aligned, the Cloud copy differed in only two places:

    • The processor note, which applies on-prem only. Cloud always uses Flink.
    • The Active-Active bullet. Active-Active can't be an RDI Cloud target, but on-prem RDI can target it.

    The embed-md shortcode can't vary its output per page. So I split the embed into rdi-when-to-use.md and rdi-when-not-to-use.md, and the on-prem note sits between them.

  • Cloud Active-Active limit: moved into the Prerequisites limitations note, with the other target-database rules: "The target database can't use Active-Active topology."

  • 1.19.1 archive: its when-to-use page had used the same embeds. I inlined its current content (classic numbers, old note, old decision tree) so the frozen snapshot doesn't change.

For review

  • @mirko: the Cloud "When not to use" bullet now says "replicaOf and Active-Active" are other ways to replicate data across Redis databases, the same wording as on-prem.
  • The shared bullet "You want a self-managed solution or AWS based solution" still reads oddly on the Cloud page. Both copies had it before this PR, so I didn't change it here.

Verification

  • Ran a Hugo build: no new relref warnings.
  • Checked the rendered /integrate/redis-data-integration/when-to-use/, /integrate/redis-data-integration/1.19.1/when-to-use/ and /operate/rc/rdi/ pages:
    • the right numbers appear on each page
    • the note is in the right place
    • the decision tree renders
    • both H3s show in the table of contents
  • The Cloud page's .md and .json outputs include the shared lists.

🤖 Generated with Claude Code


Note

Low Risk
Documentation-only changes to RDI guidance, embed structure, and archived page content; no application or security impact.

Overview
Updates RDI “when to use” guidance so throughput and data-size thresholds match the Flink processor (default from RDI 2.0.0): 20K source changes/s, 60K full-sync and 20K CDC records/s (1KB records), and 200GB total data, including the interactive decision tree questions.

Refactors content reuse: splits the former single embed into rdi-when-to-use.md and rdi-when-not-to-use.md, which both the integrate and Redis Cloud pages include; the integrate page adds a note that limits apply to Flink (classic is ~half). Redis Cloud drops duplicated prose in favor of those embeds and moves the Active-Active target restriction into the prerequisites limitations list.

The 1.19.1 archived page inlines the prior classic limits, note, and decision tree so that frozen version doc does not pick up the new numbers.

Reviewed by Cursor Bugbot for commit f482d15. Bugbot is set up for automated code reviews on this repo. Configure here.

@github-actions

github-actions Bot commented Oct 1, 2026 •

Copy link
Copy Markdown
Contributor

DOC-7142

@github-actions

github-actions Bot commented Oct 1, 2026 •

Copy link
Copy Markdown
Contributor

🧠 Redis Memory

Found 8 related items from repository history (3 new this commit):

Memory updated at f482d15

@andy-stark-redis andy-stark-redis self-assigned this Oct 1, 2026
@andy-stark-redis
andy-stark-redis requested a review from a team October 1, 2026 12:44

@dwdougherty dwdougherty left a comment

Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Just a few relrefs for you to consider converting.

Comment on lines +52 to +53
into a [nested Redis JSON object]({{< relref "/integrate/redis-data-integration/data-pipelines/data-denormalization#joining-one-to-many-relationships" >}}).
- RDI supports the [data transformations]({{< relref "/integrate/redis-data-integration/data-pipelines/transform-examples" >}}) you need for your app.

Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Convert relrefs to ... you know ... the new thing.

confirmed that they are acceptable.

{{< note >}}The throughput and data-size limits above assume the
[classic processor]({{< relref "/integrate/redis-data-integration/architecture/classic-vs-flink" >}}).

Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This one too.

{{< embed-md "rdi-when-to-use.md" >}}

{{< note >}}The throughput and data-size limits above apply to the
[Flink processor]({{< relref "/integrate/redis-data-integration/architecture/classic-vs-flink" >}}),

Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

And this one.

@andy-stark-redis

Copy link
Copy Markdown
Contributor Author

@dwdougherty Thanks for reminding me about that. RDI got missed out from the original batch of relref conversions. There were some nuisance source files that related to a work in progress for RDI dashboards. However, this has now been cancelled, so I can go ahead and fix the RDI shortcodes. These current old-style relrefs in this PR are OK for now - they'll all get converted along with the others when I do the bulk conversion.

andy-stark-redis and others added 2 commits October 2, 2026 10:20
… Cloud and on-prem lists

RDI 2.0.0 made the Flink processor the default, so the on-prem "When to use
RDI" list now quotes the Flink limits (20K changes/s, 60K full sync, 20K CDC,
200GB) that the Redis Cloud page already used, and the note is reworded to
say the limits are Flink's and the classic processor supports about half.
The decision tree thresholds (10K/100GB) moved to 20K/200GB with them so the
tree and list on one page agree.

The Cloud page had its own inline copy that drifted from the embed. With the
numbers aligned, the copies differed only in the processor note (on-prem
only; Cloud always runs Flink) and one Cloud-only bullet saying Active-Active
can't be an RDI Cloud target (on-prem can target AA). embed-md takes no
parameters and renders the embed page's .Content, so an embed can't vary per
includer. Instead the embed is split into rdi-when-to-use.md and
rdi-when-not-to-use.md so the on-prem page can put its note between them, and
the Cloud AA restriction moved into the Prerequisites limitations note beside
the other target-database rules.

The archived 1.19.1 when-to-use page used the same two embeds, so editing
them would have silently rewritten the frozen snapshot (classic was still the
default in 1.19.1). Its content is now inlined with the old numbers, old note
and old decision tree.

Learned: shared embeds can't vary per page; versioned archive pages that embed them change whenever the embed does
Constraint: rdi-when-to-use.md and rdi-when-not-to-use.md are shared by on-prem and Cloud; keep platform-specific content out of them
Rejected: Cloud-only bullet appended after the when-not embed | renders as a second <ul> with a visible gap
Directive: put Cloud-only RDI limits in rc/rdi/_index.md's Prerequisites note, not in the shared embeds
Recheck: if RDI publishes new Flink throughput benchmarks, update both the embed and the decision tree
Ticket: DOC-7142
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
… hooks

Review on #4176 flagged relrefs that the first commit brought into two pages
DOC-7086 had already converted: the new Flink note, and the embed text
inlined into the 1.19.1 page. Converted both pages with
build/migrate_shortcode_links.py all.

On the 1.19.1 page, the inlined links first pointed at the latest pages,
because the embed they came from always rendered that way. They now point at
the 1.19.1 copies, matching how other 1.19.1 pages link to their siblings.
All five targets and anchors exist under 1.19.1/.

Constraint: links in 1.19.1/when-to-use.md target 1.19.1/ pages, not latest
Ticket: DOC-7142
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@andy-stark-redis
andy-stark-redis force-pushed the DOC-7142-update-rdi-documentation-for-flink-default-in-version-2-0-0 branch from a130a6d to f482d15 Compare October 2, 2026 09:28
@andy-stark-redis
andy-stark-redis merged commit f687a7a into main Oct 2, 2026
52 checks passed
@andy-stark-redis
andy-stark-redis deleted the DOC-7142-update-rdi-documentation-for-flink-default-in-version-2-0-0 branch October 2, 2026 09:31
@andy-stark-redis

Copy link
Copy Markdown
Contributor Author

Follow-up to my earlier reply: I converted the relrefs in this PR in the end, so they didn't wait for the bulk pass. Both when-to-use pages now use render-hook links and notes. On the 1.19.1 page, the links copied in from the embed now point at the 1.19.1 copies of their targets, the same as the other 1.19.1 pages. The relrefs still left in RDI are in rc/rdi/ and the shared content/embeds/rdi-* files. I'm converting those next as DOC-7118 unit 7.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants