[APM] AWS lambda estimated cost#143986
Conversation
cauemarcondes
added
release_note:skip
|
Pinging @elastic/apm-ui (Team:APM) |
| faasTriggerTypes: response.aggregations?.faasTriggerTypes.buckets.map( | ||
| (bucket) => bucket.key as string | ||
| ), | ||
| hostArchitecture: host?.architecture, |
There was a problem hiding this comment.
is it possible that a service could have N lambda functions with different architectures?
in that case, which architecture do we display in the popup?
There was a problem hiding this comment.
Yes, it is possible. We agreed to show the most recent one so I'm going to sort the result.
There was a problem hiding this comment.
Can you please elaborate why the recent one cause it's a little bit confusing to me 🙈, usually in the popup we display summary metadata for the service.
when I see the following:
I expect both functions have arm architecture. (but it can be different)
I think, for the popup can be solved with simple terms agg on the host.architecture.
There was a problem hiding this comment.
I'm afraid if we show both the user might be confused about the estimated cost. As we use the most recent one there. @AlexanderWert WDYT? should we show a list of architecture here or just the most recent?
There was a problem hiding this comment.
@caue The point about the recent one was in the case that someone changes the architecture of a specific function (even with APM services that have only a single function associated).
But if we have an APM service with multiple functions, I think we should show both here.
TBH I haven't thought about the cost estimation for a service that has multiple different functions (with different architectures and different memory sizes).
I think, in this case we would need to calculate the cost for each function individually and then sum up the result. Not only because of the architecture but also because they might have different memory sizes, which also influences the costs significantly. Not sure if we are doing this here already or how complicated that would be?
But I think for this first iteration it's also fine to just keep it simple and show an estimate based on the recent values. (The default setup is anyways a 1:1 mapping between function and APM service)
There was a problem hiding this comment.
Agreed with @AlexanderWert that I'll try to tackle this on another PR.
| titleSize="s" | ||
| description={i18n.translate( | ||
| 'xpack.apm.serverlessMetrics.summary.estimatedCost', | ||
| { defaultMessage: 'Estimated costs avg.' } |
There was a problem hiding this comment.
I think it would be useful to have a tooltip to inform the user how the avg cost is calculated
There was a problem hiding this comment.
@boriskirov can you come up with a copy, please?
x-pack/plugins/apm/server/routes/metrics/serverless/get_serverless_summary.ts
Outdated
Show resolved
Hide resolved
x-pack/plugins/apm/server/routes/metrics/serverless/get_serverless_summary.ts
Outdated
Show resolved
Hide resolved
There was a problem hiding this comment.
Thanks for addressing most of the feedback. 🙏 Overall the code looks good to me 💯
I believe the tooltip and the discussion around the architecture in the popup don't block this PR so It's up to you in case you'd like to address them in a follow up PR :D
Thanks 🥇
x-pack/plugins/apm/server/routes/metrics/serverless/get_serverless_summary.ts
Outdated
Show resolved
Hide resolved
x-pack/plugins/apm/server/routes/metrics/serverless/get_serverless_summary.ts
Outdated
Show resolved
Hide resolved
x-pack/plugins/apm/server/routes/metrics/serverless/get_serverless_summary.ts
Outdated
Show resolved
Hide resolved
x-pack/plugins/apm/server/routes/metrics/serverless/helper.test.ts
Outdated
Show resolved
Hide resolved
| const architecture = 'x86_64'; | ||
| it('returns correct cost when usage is less than 6 b gb-sec', () => { | ||
| expect( | ||
| calcEstimatedCost({ |
There was a problem hiding this comment.
I think it'd be useful to have several invocations here, it would make it more clear from just reading the test how different factors influence the final result.
| uiSettingsClient | ||
| .get<string>(apmAWSLambdaPriceFactor) | ||
| .then( | ||
| (value): AWSLambdaPriceFactor => | ||
| JSON.parse(value) as AWSLambdaPriceFactor | ||
| ), | ||
| uiSettingsClient.get<number>(apmAWSLambdaRequestCostPerMillion), |
There was a problem hiding this comment.
Is there no way to get multiple settings from the uiSettingsClient in one call? what does it do under the hood? does each call trigger an Elasticsearch request, or is it returned from a local cache?
There was a problem hiding this comment.
There's a .getAll() function too. Under the hood .get(key) calls getAll but only returns the key received. And yes, the value is cached.
x-pack/plugins/apm/server/routes/metrics/serverless/get_serverless_summary.ts
Outdated
Show resolved
Hide resolved
💚 Build Succeeded
Metrics [docs]Public APIs missing comments
Any counts in public APIs
Async chunks
Page load bundle
Unknown metric groupsAPI count
ESLint disabled in files
ESLint disabled line counts
Total ESLint disabled count
History
To update your PR or re-run it, just comment with: |
* main: (43 commits) [Synthetics] Step details page screenshot (elastic#143452) [Lens] Datatable expression types improvement. (elastic#144173) [packages/kbn-journeys] start apm after browser start and stop after browser is closed (elastic#144267) [Files] Make files namespace agnostic (elastic#144019) Implement base browser-side logging system (elastic#144107) Correct wrong multiplier for byte conversion (elastic#143751) [Monaco] Add JSON syntax support to the Monaco editor (elastic#143739) CCS Smoke Test for Remote Clusters and Index Management (elastic#142423) [api-docs] Daily api_docs build (elastic#144294) chore(NA): include progress on Bazel tasks (elastic#144275) [RAM] Allow users to see event logs from all spaces they have access to (elastic#140449) [APM] Show recommended minimum size when going below 5 minutes (elastic#144170) [typecheck] delete temporary target_types dirs in packages (elastic#144271) [Security Solution][Endpoint] adds new alert loading utility and un-skip FTR test for endpoint (elastic#144133) [performance/journeys] revert data_stress_test_lens.ts journey step (elastic#144261) [TIP] Use search strategies in Threat Intelligence (elastic#143267) Optimize react-query dependencies (elastic#144206) [babel/node] invalidate cache when synth pkg map is updated (elastic#144258) [APM] AWS lambda estimated cost (elastic#143986) [Maps] layer group wizard (elastic#144129) ...
8 participants
Part of #142328