CCS Components: State Storage Deep Dive

CCS Components: State Storage Deep Dive Agenda

• Timing
  • 15-20 min: Ulf Intro presentation on CSS Components: State Storage.
  • 40 min: reaction and building future plan
• Technical API discussion
  • Design Issues in UIf's presentation
  • Review current State Storage component API
    
    • Anything to fix or extend?
    • Categorise in wider architecture:
      • In VSS Terminology State Storage is 'simple'. Is this component strong coupled (fit) for that use case and something else is needed for 'more powerful' or do we reuse/extend?
      • Simple vs powerful may have multiple dimensions: throughput, features (e.g. events), etc.
  • Related areas (likely will need their own Deep Dive sessions as too big a topic to conclude here)
    • Abstraction APIs
      • To Feeder, to Server
      • Requirements
        • Speed
          • Low vs medium vs high frequency
          • Native for high frequency?
      • Kuksa.val is also a target
    • Last value vs time series
      • VISS protocol has some TS query
        • 'History filtering'
        • Simple support for this in current WAII. Other process (e.g. comms) must tell WAII to start/stop recordering. Then VISS client (e.g. cloud) must detect vehicle became disconnected and query for missing data.
        • Kuksa currently has no TS
• Date
  • Monday 23rd May 4pm CET.

• Rough meeting notes:
  • Attendees: Ted Guild, Ulf, Paul, James Murphy, Florian Pinzel, Stephen, Jose Gomez
  • Presentation: COVESA AMM 2022 - State storage [PUB].pdf
    • Origin of component in CCS Project
    • Data structure
      • Actuator model encapsulated (abstracted) in State Storage component
      • VSS Path list
        • SQLite has manager to help create path list.
        • Redis has no need for path list.
    • Design Issues
      • Polling vs event notification
        • Ulf: looked into SQLite event framework but it didn't seem to fit the needs here. Not looked into Redit yet.
        • Discussion of requirements of slow to medium frequency data vs high frequency
  • Futures
    • Timeseries, event notification, app framework etc.