Cloud app design and dev

Cloud app design and dev

1. Design

   • Creating a product, service, or other plan to satisfy a specific end goal
   • Bad design
     • Slow response
     • Application outages
     • Vulnerable application
     • Inefficient usage of resources
     • Higher cost
     • Lacks scalability

2. Cloud and traditional different

   • Underlying Infrastructure may Reboot
     • computing resources may be go down and restart at anytime
     • get replaced by a different machine altogether
     • includes the entire stack of infrastructure, middleware and service
     • cannot store data on the local drive on the cloud server
     • Cloud charicaristic
       • shared computing resources
       • management software pulls together
   • infinitely scalable
     • ability to pull resources to scale
     • horizontal scaling, not vertical
     • adding extra instance or same thing, and load balance
     • application has to be designed to cater for this behaviour
   • Everything is online
     • always connected
     • maybe used 24x7
     • need to find suitable maintenance window
     • Security consideration
     • API - risk for hackers
     • two level of security
       • Application level
       • Network level

3. key architecture

   • design to scale horizontally
     • scale by adding more app/DB instances
     • use load balance
     • auto-scaling
   • Don't persist data on local server
     • cloud server and service may reboot any time
     • local server storage is not reliable
     • store session data on cloud-based caching facilities
     • store persisting data on cloud database or BLOBs
   • Explicitly declare and Isolate dependencies
     • never assume external data exists, declare dependencies and automatical download of library
     • make sure put in procedure to handle error from web services
   • Componentise an application
     • break app down into small building blocks
     • blocks work with each other to fulfil app function
     • faster deployment as deploy a small part each time
     • Microservice architecture
   • Loosely Couple the components
     • components are encapsulated
       • communication relies on sending and receiving messages via standard interface exposed to other components
     • changing internal logic for one do not require change of another
     • Microservice architectur
     • message queue or message broker technologies
       • ensure reliable delivery of message
   • Design stateless App
     • stateful app store state info in app server
     • Cloud app run on many server
     • Desing app that state info defined by DB and shared storage.
   • Consider security
     • Network layer
       • Firewall
       • Intrusion Detection System
       • Intrusion Prevention Sys
       • System Incident Event Management
     • Application layer
       • Authentication and authorisation
       • security built into forms to prevent SQL injection
       • encryption and masking DB, consider needed data
       • encryption of data in transit
       • manual or automated penetration
   • Cost and performance
     • minimise calls to APIs
   • Use Micro-services
     • modular applications that can communicate with each other and work in tandem.
     • Characteristics
       • Componentisation via services
       • Organised around Business Capabilities
       • Products not Projects
       • Smart endpoint and dumb pipes
       • Decentralised Governance
       • Decentralised Data Management
       • Infrastructure Automation
       • Design for failure
       • Evolutionary Design
     • Benefits
       • Failing of 1 service will not affect other
       • can be upgraded individually
       • increase modularity - easier to understand
       • fast to deploy and start
       • allowing team to try multiple technologies
     • 3-dimension scalability model
       • Y-axis - split by functionality
         • by none or by verb
       • x-axis -- clone and load balance
       • z-axis - partition into subsets (more on DB)
     • Challenges
       • balance the granularity of service
       • Tracking service dependency
       • Fault tolerance architecture
       • Resulting in lots of traffic
       • Decentralised data management
       • Coordinating deployment across multiple micro-services

4. Design patterns

   • proven solutions for repeated problems
   • saves time and cost, decreases training time
   • encourage better documentation and community support
   • incorporates a number of design principles