Skip to content
/ optima Public

Distributed Dataset management and Calculation service

Notifications You must be signed in to change notification settings

cotyar/optima

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

52 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Optima

Distributed Dataset management and Calculation service

Dealing with datasets naturally


Entities

  • Datasets
  • Transformations

Dataset

  • Is identifiable
  • Can be versioned
  • Can be Chunked
  • Of type
    • Immutable
      • Id is Hash (a checksum based on a cryptographic hash) digest
        • SipHash-2-4 (64bit, fast and "good enough")
        • MD5
        • SHA256 (strongest and longest)
      • Is idemportent
    • Streamed
      • Id is globaly unique (UUID)
      • Can support autobatching into chunks (do we need this?)
        • Chunks are organized in a MerkleTree
        • Not necessary idemportent
          • head chunk(s) can be dropped
          • tail chunk is naturally mutable
      • Can simply wrap Kafka/Debezium
  • Is named
  • Is owned
  • Can be groupped
  • Has RBAC
  • Is searchable
  • Is audited
  • Has a strongly verifiable format (Schema)
  • Has heritage tracked (lineage)
  • Can be represented by:
    • File
      • text
      • binary
      • csv
      • json
      • parquet
    • Database table
      • SQL
        • Oracle
        • PostgreSql
        • Sqlite
      • KV
        • RocksDB
          • Protobufs
  • Has a streaming endpoint
    • gRPC (mandatory)
    • Thrift (optional)
  • Governance
    • in Git following GitOps principles
    • rows as .proto messages

Transformation

  • Can be one-off
  • Can be executed
  • Can be reoccuring
  • Can be re-executed
    • creating a new Dataset
    • creating a new version of existing Dataset
  • Has RBAC
    • Run
    • RerunFailed
  • Can be monitored
  • Governance
    • in Git following GitOps principles
    • rows as .proto messages
  • Data masking (optional)
    • role sensitive masking

Infrastructure

  • Sensible defaults
  • Implemented:
    • with Dapr
      • MQ agnostic
      • Storage agnostic
      • Cloud agnostic
      • Database agnostic
      • does Virtual Actors
      • is simply brilliant
    • as Service Mesh
      • Linkerd or Istio
    • in K8s
    • in Docker containers
  • Governance
    • in Git following GitOps principles
      • Fluxcd
    • usage quotas (optional)
  • Logging
    • Logz
    • Zipkin
    • AppInsights
    • Jaeger
    • etc
  • Location awarness (optional)

Monitoring

  • gRPC liveness probe
  • K8s monitoring tools
    • Prometeus
    • Grafana
    • Linkerd, Flux, etc. admin pages

Principles

  • Data masking out-of-the-box
  • "The cheapest way of doing something is not doing it at all"
  • "Staying on the shoulders of giants"
  • "nocode" - "Don't solve in code what can be solved in the infrastructure"
  • Automate, automate, automate
  • k8s are capable to doing a lot

About

Distributed Dataset management and Calculation service

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published