redhat-developer / redhat-developer/mapt
[enhancement] Spot heuristics module
Nobody has claimed this yet.
- Dominant language
- Go
- Stars
- 24
- Forks
- 36
- Avg merge
- 6d 13h
- Merged PRs (30d)
- 8
Description
There are some situations where we can not use data for calculate odds:
-
Recently we found out due to #593 some (all?) opt-in regions from AWS do not offer SpotPlacementScore data, can not find if that will be supported at any time. As so those regions are directly discarded from spot calculations.
-
On top of this, from time to time we also miss Eviction metric from Azure due to some unknown issues on their graphql service. In that case we use directly price info.
-
Also on some previous research other providers do not support such metric.
This issue proposes to create a generic heuristic module to emulate the score so it can be used on those cases were we will not be able to retrieve that data:
Actual Data
The proposed solution will calculate the score based on the formula
discount = 1 - (spot_price / on_demand_price)
Temporal data
When possible data formula could be improved with temporal data:
Compute deltas or ratios:
delta1d = (current - yesterday) / yesterday
delta7d = (current - lastWeek) / lastWeek
Use those deltas as extra signals:
If the current price is much lower than historical → maybe surplus capacity (good).
If much higher or spiking vs history → maybe risk of interruption (bad).
@anjannath WDYT?
Contributor guide
No contributing guide indexed for this repository
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
No files, tests, or entry points are named. Start by reviewing the existing AWS SpotPlacementScore and Azure eviction/price handling, along with issue #593, then determine where a provider-neutral heuristic belongs. Done should mean the proposed discount formula can cover cases where spot metrics are unavailable; temporal signals remain an optional design point.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- aws, azure, go
- Domain
- cloud
- Issue type
- Feature
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Stale
- Clarity
- Mostly clear
- Newbie friendliness
- 30/100