Nearby-backend

Geospatial based photo shared social network backend with Golang
Frontend is implementated with ReactJS, please see here.

Overview of project

• Develop Web services with Golang to handle user auth(register, login, logout), post and search
• Implement token based authentication with PostgreSQL, JWT, OAuth 2.0
• Implement posts storage, geo-location based search for nearby posts, data visualization with ElasticSearch and Kibana in GKE(Google Kubernetes Engine)
• Store post image with Google Cloud Storage(GCS)
• Improve read performance with a little data consistency sacrifice with Redis
• Deploy with Docker, GKE(Google Kubernetes Engine) and GOOGLE CLOUD LOAD BALANCING

API

• /signup

  • save to PostgreSQL.

• /login

  • check login credential in PostgreSQL, if correct return token

• /search - search nearby posts

  1. have token-based authentication first
  2. search in Redis cache, if not found then search ElasticSearch
  3. use "type" : "geo_point" to map (lat, lon) to geo_point, ElasticSearch will use geo-indexing to search(KD tree)

  e.g.

  /search?lat=10.0&lon=20.0
  /search?lat=10.0&lon=20.0&range=10
  

• /post

  1. save post image in GCS.
  2. save post info in ElasticSearch, bigTable(optional).

  post body

  {
    "message": "<3",
    "location": {
      "lat": 47.651977307189256,
      "lon": -122.3316657342046
    },
    "url": [
      "https://www.googleapis.com/download/storage/v1/b/nearby-posts/o/d46f6e06-669b-41ec-9bb4-25a6e08c259f0?generation=1548962551529115&alt=media"
    ]
  }
  

TODO

• CRUD for user's own posts
• Follow and Like