Repository navigation
Conversation
Signed-off-by: zerodrift <zerodrift@users.noreply.github.com>
This branch has not been deployed
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Summary
This experiment builds on #321. It compares two ways to display a synthetic power curve. The control prices all deliveries together. The progressive version calculates and streams the first eight prices, then calculates the remaining deliveries against the same pinned snapshot. Both use gzip JSON over SSE and the workbench chart renderer.
Results
Three independent Release runs produced 24 warm observations per workload and mode. The table gives median times with 330 ms of added round-trip latency.
Recommendation: keep the compact whole-curve read in the application. Streaming brought the first prices forward by 14–50 ms, below the 100 ms target. Splitting the build repeats snapshot validation and plan setup, and both workloads took longer to finish. The trial starts from an in-memory model and excludes database access, first-use hydration, authorization and the deployed reverse proxy.
Verification
The test runs in an opt-in executable on loopback. Its workload, p50/p95 results and reproduction command are in the verification record.