Please enable JavaScript.

Coggle requires JavaScript to display documents.

Big Data Storage Concepts - Coggle Diagram

- - - - to minimize downtime and provide uninterrupted service when nodes fail.
      - highly fault tolerant
      - • Nodes in the cluster continuously communicate and share information (e.g., through heartbeat signals
      - Fault Tolerance & Redundancy:
      - • Elimination of Single Point of Failure:
    - - • Objectives:
        
        • Maximize throughput.
        
        • Minimize response time.
        
        • Optimize resource usage.
        
        • Avoid overload on any one node.
        
        • Efficient Resource Use:
      - • Load Balancing Algorithms / Techniques:
        
        Weight-Based:
        
        Random:
        
        Round Robin:
        
        Server Affinity:
  - - -  All nodes are active and run applications.
      -  Workload and monitoring are shared.
      -  No node is purely passive (like a “standby” that does nothing until failure).
    - -  Some nodes are designated to be “hot standby” or passive; others are active.
      -  The passive node(s) mainly monitor for failure but do not normally handle workload until needed.
      -  When an active node fails, a standby takes over.
- - - - fault tolerant  the data is not lost when an individual node fails as the data is redundant across the nodes.
      - Replication increases the data availability as the same copy of data is available across multiple nodes.
    - - peer‐peer models
      - master‐slave
  - - - • Scalability:
      - • Performance:
      - • Manageability:
      - • Fault Tolerance:
      -  Failure of one node only affects its shard, not the entire dataset.
- - - - • Schema-based:
      - • Limitation:
      - • ACID Properties:
        → reliable transactions.
        
        Consistency
        
        Isolation
        
        Atomicity
        
        Durability
  - - - • Schemaless Design:
      - • Supports Multiple Data Formats:
      - • Ideal for Big Data:
      - • CAP Theorem
        
        Consistency
        
        Availability
        
        • Partition Tolerance:
      - • BASE Model:
        
        Soft state, and Eventually consistent
        
        Available
        
        Basically
    - - • Combine NoSQL scalability with ACID properties of traditional RDBMS.
      - • Examples: VoltDB, NuoDB, Clustrix, MemSQL, TokuDB.
      - • Fault tolerant
      - • Data Model:
        
        Support relational model
        
         Provide SQL-compliant syntax (easy transition from RDBMS).
      - • Applications:
        
        Systems that execute the same queries repeatedly with different inputs
        
         Workloads with high transaction volumes
      - • Distributed architecture