Skip to content

[WIP] Add blog post: Benchmarking AI on a simulated KubeEdge cluster with Ianvs - #890

Open
PragnyaKhandelwal wants to merge 6 commits into
kubeedge:masterfrom
PragnyaKhandelwal:blog/ianvs-simulation-sandbox
Open

[WIP] Add blog post: Benchmarking AI on a simulated KubeEdge cluster with Ianvs#890
PragnyaKhandelwal wants to merge 6 commits into
kubeedge:masterfrom
PragnyaKhandelwal:blog/ianvs-simulation-sandbox

Conversation

@PragnyaKhandelwal

@PragnyaKhandelwal PragnyaKhandelwal commented Sep 30, 2026 •
edited
Loading

Copy link
Copy Markdown
  • Please check if the PR fulfills these requirements
  • The commit message follows our guidelines
  • Tests for the changes have been added (for bug fixes / features)
  • Docs have been added / updated (for bug fixes / features)

Which issue(s) this PR fixes:

None. This is a new blog post for the CNCF LFX mentorship project on the KubeEdge-Ianvs Simulation Sandbox (proposal, tracked in kubeedge/ianvs#348).

  • What kind of change does this PR introduce? (Bug fix, feature, docs update, ...)

Blog post.

  • What is the current behavior? (You can also link to an open issue here)

There is no post about benchmarking AI on a simulated edge-cloud cluster with KubeEdge-Ianvs.

  • What is the new behavior (if this is a feature change)?

Adds blog/ianvs-simulation-sandbox/index.mdx, a post on:

  • why a simulator matters, in time and money: a before/after table with measured numbers (a ready cluster with an edge node in 4 min 16 s, a model that does not fit a 1 GiB device caught in 13 s, about 1.7 GB of RAM at the peak, cleanup in 6 s, $0 of extra hardware);
  • who it is for: algorithm developers, researchers, contributors and maintainers, students and newcomers;
  • how the sandbox works: Ianvs builds a kind + KubeEdge + Sedna cluster on one Linux host and runs each test case as a Kubernetes Job on a simulated edge node, with the memory and CPU limits of a device;
  • the work behind it: the restored simulator, the Sedna all-in-one update (Kubernetes v1.30.0, KubeEdge v1.21.0, Sedna v0.7.0) and the Simulation Job Administrator;
  • the test environment, the measured runs, how to try it, and the current limits.

Pictures and the demo GIF are at the top and the tutorial is short with a link to the Ianvs guide, following the feedback on #752.

Also adds the author to blog/authors.yml and i18n/zh/docusaurus-plugin-content-blog/authors.yml.

  • Does this PR introduce a breaking change? (What changes might users need to make in their application due to this PR?)

No.

  • Other information:

The numbers in the post are measured on one laptop; the recorded runs are in kubeedge/ianvs#1012 and kubeedge/ianvs#1041. The hardware prices are linked in the post. The post says that the work is in review (kubeedge/ianvs#1012, kubeedge/ianvs#1041, kubeedge/sedna#513); I will update that paragraph when they merge.

@kubeedge-bot kubeedge-bot added the do-not-merge/work-in-progress Indicates that a PR should not merge because it is a work in progress. label Sep 30, 2026
@kubeedge-bot

kubeedge-bot commented Sep 30, 2026

Copy link
Copy Markdown
Collaborator

[APPROVALNOTIFIER] This PR is NOT APPROVED

This pull-request has been approved by:
To complete the pull request process, please assign wbc6080 after the PR has been reviewed.
You can assign the PR to them by writing /assign @wbc6080 in a comment when ready.

The full list of commands accepted by this bot can be found here.

Details Needs approval from an approver in each of these files:

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment

@kubeedge-bot

kubeedge-bot commented Sep 30, 2026

Copy link
Copy Markdown
Collaborator

Welcome @PragnyaKhandelwal! It looks like this is your first PR to kubeedge/website 🎉

@kubeedge-bot kubeedge-bot added the size/L Denotes a PR that changes 100-499 lines, ignoring generated files. label Sep 30, 2026
@PragnyaKhandelwal

PragnyaKhandelwal commented Sep 30, 2026

Copy link
Copy Markdown
Author

/hold

@kubeedge-bot kubeedge-bot added the do-not-merge/hold Indicates that a PR should not merge because someone has issued a /hold command. label Sep 30, 2026
PragnyaKhandelwal added a commit to PragnyaKhandelwal/ianvs that referenced this pull request Oct 3, 2026
@PragnyaKhandelwal
docs(proposal): simulation sandbox metrics, glossary and docs plan
06fd4b6 
Address review feedback on the Ianvs Simulation Sandbox proposal.

- §6.3: correct the config contract to the existing simulation: key
  (cloud_number, edge_number, ...); add system_metrics: "all" and the
  optional device_simulation block (max_memory_mb, max_cpu_cores,
  network_bandwidth_mbps, network_latency_ms) with network_profile.
- §6.4: system metrics defined at the paradigm level for all seven
  paradigm types, grouped by technical dependency; add GPU (utilization,
  memory, power, energy) and storage metrics (disk usage, I/O, model
  artifact size), federated data and exchange metrics, the relationship
  to the existing SystemMetricType metrics, and a compact metric
  glossary with worked examples.
- §6.11: paradigm-aware device_simulation defaults for all seven types.
- §2, §6.3, §7: pin the KubeEdge default to v1.14.0, the newest version
  verified end-to-end, and explain why newer versions fail with the
  Sedna all-in-one installer (the keadm v1.9.1 in its node image
  requests v1alpha2 device CRDs that are gone from release-1.16 on, and
  the image runs Kubernetes 1.21).
- §9.2: parallel-processing notes for federated class-incremental
  learning and multi-edge inference.
- §10: documentation plan for the simulation README, the KubeEdge blog
  post (kubeedge/website#890) and the Linux Foundation mentorship blog.

Design only; no paradigm internals change.

Signed-off-by: Pragnya Khandelwal <prag1704@gmail.com>
PragnyaKhandelwal added a commit to PragnyaKhandelwal/ianvs that referenced this pull request Oct 3, 2026
@PragnyaKhandelwal
docs(proposal): simulation sandbox metrics, glossary and docs plan
4f0e4c5 
Address review feedback on the Ianvs Simulation Sandbox proposal.

- §6.3: correct the config contract to the existing simulation: key
  (cloud_number, edge_number, ...); add system_metrics: "all" and the
  optional device_simulation block (max_memory_mb, max_cpu_cores,
  network_bandwidth_mbps, network_latency_ms) with network_profile.
- §6.4: system metrics defined at the paradigm level for all seven
  paradigm types, grouped by technical dependency; add GPU and storage
  metrics, federated data and exchange metrics, the relationship to the
  existing SystemMetricType metrics, and a compact metric glossary with
  worked examples.
- §6.11: paradigm-aware device_simulation defaults for all seven types.
- §2, §6.1, §6.6, §6.8, §7: align Stage 1 with the restoration in
  kubeedge#1012 (no auto-install, cgroup-aware checks, aliases for misspelled
  names, KubeEdge pinned to v1.14.0 with the reason newer versions fail
  with the Sedna all-in-one installer).
- §6.2, §6.8, §6.9, §6.10: test cases run as Kubernetes Jobs on
  simulated edge nodes (Sedna has no LearningJob CRD); fix the example
  paths and leaderboard columns in the user flow.
- §9.2: parallel-processing notes for federated class-incremental
  learning and multi-edge inference.
- §10: documentation plan for the simulation README, the KubeEdge blog
  post (kubeedge/website#890) and the Linux Foundation mentorship blog.

Design only; no paradigm internals change.

Signed-off-by: Pragnya Khandelwal <prag1704@gmail.com>
PragnyaKhandelwal added a commit to PragnyaKhandelwal/ianvs that referenced this pull request Oct 8, 2026
@PragnyaKhandelwal
docs(proposal): simulation sandbox metrics, glossary and docs plan
6f079e7 
Address review feedback on the Ianvs Simulation Sandbox proposal.

- §6.3: correct the config contract to the existing simulation: key
  (cloud_number, edge_number, ...); add system_metrics: "all" and the
  optional device_simulation block (max_memory_mb, max_cpu_cores,
  network_bandwidth_mbps, network_latency_ms) with network_profile.
- §6.4: system metrics defined at the paradigm level for all seven
  paradigm types, grouped by technical dependency; add GPU and storage
  metrics, federated data and exchange metrics, the relationship to the
  existing SystemMetricType metrics, and a compact metric glossary with
  worked examples.
- §6.11: paradigm-aware device_simulation defaults for all seven types.
- §2, §6.1, §6.6, §6.8, §7: align Stage 1 with the restoration in
  kubeedge#1012 (no auto-install, cgroup-aware checks, aliases for misspelled
  names, KubeEdge pinned to v1.14.0 with the reason newer versions fail
  with the Sedna all-in-one installer).
- §6.2, §6.8, §6.9, §6.10: test cases run as Kubernetes Jobs on
  simulated edge nodes (Sedna has no LearningJob CRD); fix the example
  paths and leaderboard columns in the user flow.
- §9.2: parallel-processing notes for federated class-incremental
  learning and multi-edge inference.
- §4, §6.1, §9.5: record scalability (simulating more nodes per host,
  or across hosts) as future work, deferred per mentor guidance.
- §10: documentation plan for the simulation README, the KubeEdge blog
  post (kubeedge/website#890) and the Linux Foundation mentorship blog.

Design only; no paradigm internals change.

Signed-off-by: Pragnya Khandelwal <prag1704@gmail.com>
PragnyaKhandelwal added a commit to PragnyaKhandelwal/ianvs that referenced this pull request Oct 8, 2026
@PragnyaKhandelwal
docs(proposal): simulation sandbox metrics, glossary and docs plan
8d54d8e 
Address review feedback on the Ianvs Simulation Sandbox proposal.

- §6.3: correct the config contract to the existing simulation: key
  (cloud_number, edge_number, ...); add system_metrics: "all" and the
  optional device_simulation block (max_memory_mb, max_cpu_cores,
  network_bandwidth_mbps, network_latency_ms) with network_profile.
- §6.4: system metrics defined at the paradigm level for all seven
  paradigm types, grouped by technical dependency; add GPU and storage
  metrics, federated data and exchange metrics, the relationship to the
  existing SystemMetricType metrics, and a compact metric glossary with
  worked examples.
- §6.11: paradigm-aware device_simulation defaults for all seven types.
- §2, §6.1, §6.6, §6.8, §7: align Stage 1 with the restoration in
  kubeedge#1012 (no auto-install, cgroup-aware checks, aliases for misspelled
  names, KubeEdge pinned to v1.14.0 with the reason newer versions fail
  with the Sedna all-in-one installer).
- §6.2, §6.8, §6.9, §6.10: test cases run as Kubernetes Jobs on
  simulated edge nodes (Sedna has no LearningJob CRD); fix the example
  paths and leaderboard columns in the user flow.
- Figures 1 and 2: update the architecture and user-flow diagrams to
  match the text (simulation: block keys, Job deployment, all paradigm
  types).
- §9.2: parallel-processing notes for federated class-incremental
  learning and multi-edge inference.
- §4, §6.1, §9.5: record scalability (simulating more nodes per host,
  or across hosts) as future work, deferred per mentor guidance.
- §10: documentation plan for the simulation README, the KubeEdge blog
  post (kubeedge/website#890) and the Linux Foundation mentorship blog.

Design only; no paradigm internals change.

Signed-off-by: Pragnya Khandelwal <prag1704@gmail.com>
PragnyaKhandelwal added a commit to PragnyaKhandelwal/ianvs that referenced this pull request Oct 8, 2026
@PragnyaKhandelwal
docs(proposal): simulation sandbox metrics, glossary and docs plan
fbb3096 
Address review feedback on the Ianvs Simulation Sandbox proposal.

- §6.3: correct the config contract to the existing simulation: key
  (cloud_number, edge_number, ...); add system_metrics: "all" and the
  optional device_simulation block (max_memory_mb, max_cpu_cores,
  network_bandwidth_mbps, network_latency_ms) with network_profile.
- §6.4: system metrics defined at the paradigm level for all seven
  paradigm types, grouped by technical dependency; add GPU and storage
  metrics, federated data and exchange metrics, the relationship to the
  existing SystemMetricType metrics, and a compact metric glossary with
  worked examples.
- §6.11: paradigm-aware device_simulation defaults for all seven types.
- §2, §6.1, §6.6, §6.8, §7: align Stage 1 with the restoration in
  kubeedge#1012 (no auto-install, cgroup-aware checks, aliases for misspelled
  names, KubeEdge pinned to v1.14.0 with the reason newer versions fail
  with the Sedna all-in-one installer).
- §6.2, §6.8, §6.9, §6.10: test cases run as Kubernetes Jobs on
  simulated edge nodes (Sedna has no LearningJob CRD); fix the example
  paths and leaderboard columns in the user flow.
- Figures 1 and 2: update the architecture and user-flow diagrams to
  match the text (simulation: block keys, Job deployment, all paradigm
  types).
- §9.2: parallel-processing notes for federated class-incremental
  learning and multi-edge inference.
- §4, §6.1, §9.5: record scalability (simulating more nodes per host,
  or across hosts) as future work, deferred per mentor guidance.
- §10: documentation plan for the simulation README, the KubeEdge blog
  post (kubeedge/website#890) and the Linux Foundation mentorship blog.

Design only; no paradigm internals change.

Signed-off-by: Pragnya Khandelwal <prag1704@gmail.com>
@PragnyaKhandelwal
PragnyaKhandelwal force-pushed the blog/ianvs-simulation-sandbox branch 2 times, most recently from 9a6a358 to 2f5a3a2 Compare October 22, 2026 07:11
@PragnyaKhandelwal
blog: add KubeEdge-Ianvs simulation post
9755885 
Work-in-progress blog post on the KubeEdge-Ianvs simulator. This revision covers the restoration: building a kind + KubeEdge + Sedna cluster on one Linux host, running the PCB-AoI benchmark with it, and removing the cluster afterwards, with real results, screenshots and a demo GIF. Sections on Stage 2 and system metrics will be added as that work lands. Also adds the author to authors.yml (en and zh).

Signed-off-by: Pragnya Khandelwal <prag1704@gmail.com>
@PragnyaKhandelwal
blog: add hero illustration to the Ianvs simulation post
b646b6c 
Replace the opening GIF with a before/after illustration of what the
simulator saves: separate machines and manual setup versus one laptop,
one command and automatic cleanup, with measured numbers.

Signed-off-by: Pragnya Khandelwal <prag1704@gmail.com>
@PragnyaKhandelwal
blog: explain the time and cost the Ianvs simulator saves
af6bada 
Add a without/with comparison (hardware, cost, setup, version
compatibility, time to a ready cluster, cleanup) using only measured or
verified numbers, a list of everyday problems the simulator solves for
algorithm developers, and a shorter history section.

Signed-off-by: Pragnya Khandelwal <prag1704@gmail.com>
@PragnyaKhandelwal
blog: add a relatable opening and a one-command timeline
10464e2 
Open with a scenario readers recognise (a model that scores well on a
laptop but has never seen an edge-cloud cluster), and add a timeline
figure of the four steps with measured durations.

Signed-off-by: Pragnya Khandelwal <prag1704@gmail.com>
@PragnyaKhandelwal
PragnyaKhandelwal force-pushed the blog/ianvs-simulation-sandbox branch from 2f5a3a2 to 10464e2 Compare November 7, 2026 17:10
@PragnyaKhandelwal
blog: update the Ianvs simulation post with the sandbox and measured …
59cfb44 
…numbers

Rewrite the post around the simulation sandbox: what it saves in time
and money, who it is for, and how test cases run as Kubernetes Jobs on
a simulated KubeEdge edge node. Update the versions to Kubernetes
v1.30.0, KubeEdge v1.21.0 and Sedna v0.7.0, and replace the
screenshots, diagrams and demo GIF with ones from the current runs.

Signed-off-by: Pragnya Khandelwal <prag1704@gmail.com>
@PragnyaKhandelwal
blog: make the Ianvs post screenshots readable and start the demo at …
e53cbf5 
…the run

Crop the terminal screenshots to the part each one shows, start the demo GIF at the benchmark run instead of the title screen, and give the numbers table a header row.

Signed-off-by: Pragnya Khandelwal <prag1704@gmail.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@anvithks anvithks Awaiting requested review from anvithks
@Shelley-BaoYue Shelley-BaoYue Awaiting requested review from Shelley-BaoYue

Assignees

No one assigned

Labels

do-not-merge/hold Indicates that a PR should not merge because someone has issued a /hold command. do-not-merge/work-in-progress Indicates that a PR should not merge because it is a work in progress. size/L Denotes a PR that changes 100-499 lines, ignoring generated files.

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

2 participants

@PragnyaKhandelwal @kubeedge-bot