Skip to content

Latest commit

 

History

History
42 lines (36 loc) · 3.35 KB

MILESTONES.md

File metadata and controls

42 lines (36 loc) · 3.35 KB
Raw

Milestones

30-day sprint [x] (Complete!)

  • Implement CanDB APIs running on the canister manager such that the stored data is persisted and stable through upgrades. APIs include:
    • Create (Capability shown through using CanDB.update()) in the simpleDB example
    • Put/Update
    • Get
    • Delete
    • Scan (across a partition key + range/sort key combination)
  • Complete unit testing for all CanDB APIs
  • Implement Red-Black tree split algorithm (For halving a partition)

Milestone 1 [x] (Complete!)

  • Define and implement mode of inter-canister communication between the CanDB canister manager and its storage partitions (async blocking, fire-and-forget, etc.)
  • Build example demonstrating the ability to apportion cycles from the Index Canister to newly created canisters
  • Implement library functionality for the CanDB Index Canister, which controls multiple storage partitions, partition storage limits, and partition data splitting logic.
  • Auto-Scale design and initial implementation
  • Build and demo a simple application interacting directly with a CanDB instance canister manager which is controlling and has data stored in multiple storage partition canisters.

CanDB alpha goals (In progress, complete in tandem with Milestone 2)

  • Cultivate an passionate alpha community dedicated to building at scale and pushing the limits of what's possible on the IC
  • Reduce developer adoption friction by improve CanDB documentation and quickstart tooling
  • Refine existing CanDB and CanDBAdmin APIs through developer feedback
  • [] Receive cycle management feedback, leading to the appropriate API abstraction -> CycleOps
    • [] Design and implement cycle management and apportioning between the CanDB canister manager and storage partition canisters. Provide API allowing the developer to query remaining cycles at the canister manager level. -> Doing in CycleOps grant #2
  • Use developer input/feedback to help scope and prioritize new features.

Milestone 2 [] (In progress)

  • [x ] Measure and fine tune performance of CanDB APIs as the number of records and storage partitions increase
    • Define metrics, such as:
      • query/update performance
      • canister memory used
      • request throughput (number of concurrent query/update calls CanDB is able to process in a short burst of time)
    • [] Test CanDB against these metrics as the number of records/partition and number of partitions increase, ensuring that these metrics match the expected runtime and performance - i.e. proof of CanScale ;)
  • [] Identify and measure performance bottlenecks, and attempt to improve upon these bottlenecks if possible
    • [] Measure inter-canister query times and look into performance tradeoffs between direct canister calls and calls proxied through the canister manager to its partitions
    • [] Depending on the performance of the Red-Black Tree in ordering records, implement a BTree library for underlying CanDB data structures allowing partition key and range/sort key access, as well as splitting a storage partition and spinning up "n" numbers new storage partitions based on the order of the BTree.
  • [] Perform data stability testing, ensuring that data stored in CanDB is persisted through upgrades
  • [] Research and test other potential "chaos" scenarios that might affect data stability.
  • [] Add integration testing