You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Directly affects the solver. Second-order effects on job creation and resource providers.
DevOps work to deploy the new database.
Describe the changes, and how this affects/ interacts with each system
This feature adds a solver store database for job offers, resource offers, deals, match decisions, and results metadata.
Our current implementation stores data in memory. This implementation blocks us from horizontally scaling the solver because it's state can only exist in one machine's memory. Moving the data into a database is a first step towards horizontally scaling the solver and persisting state through reboots and deploys.
We will initially add a Postgres database that stores each of these items as jsonb blobs. We will add indexes for common queries, such as wallet addresses and deal IDs. Over time, we can move towards a relational model as we stabilize our data types across the protocol.
The feature has second-order effects on job creators and resource providers by persisting their offers across instances and deployments. This persistence means resource providers do not need to re-submit their resource offers and job offers will not be dropped.
The text was updated successfully, but these errors were encountered:
General Description
Add a solver database that implements the solver store interface.
Which system(s) or functionality does this affect
Directly affects the solver. Second-order effects on job creation and resource providers.
DevOps work to deploy the new database.
Describe the changes, and how this affects/ interacts with each system
This feature adds a solver store database for job offers, resource offers, deals, match decisions, and results metadata.
Our current implementation stores data in memory. This implementation blocks us from horizontally scaling the solver because it's state can only exist in one machine's memory. Moving the data into a database is a first step towards horizontally scaling the solver and persisting state through reboots and deploys.
We will initially add a Postgres database that stores each of these items as
jsonb
blobs. We will add indexes for common queries, such as wallet addresses and deal IDs. Over time, we can move towards a relational model as we stabilize our data types across the protocol.The feature has second-order effects on job creators and resource providers by persisting their offers across instances and deployments. This persistence means resource providers do not need to re-submit their resource offers and job offers will not be dropped.
The text was updated successfully, but these errors were encountered: