System Design

System Design Notes

I found these notes sometime ago, i am not the author.

1. Try to break th problem into simple modules (Top Down approach)

2. Talk about the trade-off (No soltion is perfect) Calculate the impact on system based on all the constraints and the end test cases

3. Architectural pieces/resources available

4. How these resources work together

5. Utilization and Tradeoffs

   • Consistent Hashing
   • CAP Theorem
   • Load balancing
     • Type of distribution
       1. Random
       2. Round-robin
       3. Weight for CPU & Memory cycle
     • To utilize full scalabillity & redundancy(Add 3 LB[3-Tier Architecutre])
       1. User « LB1 » Web Server
       2. Web Server « LB2 » App Server / Cache Server
       3. App Server / Cache Server « LB3 » DB
     • Smart Clients
       • Takes a pool of service hosts & balances load
         • detects hosts that are not responsive
         • recovered hosts
         • addition of new hosts
       • Load balancing functionality to DB(Cache service)
       • Attractive solution for developers
       • As system grows > LBs (Standalone Servers)
     • Hardware Load Balancers
       • Expensive but high preformance
       • Not trivial to configure
       • Large companies tond to avoid this config or use it as 1st point of contact to their system to serve user requests
       • Intra network uses smart clients / hybrid solution for load balancing traffic
     • Software Load Balancers
       • No pain of creation of smart client
       • No cost of purchasing dedicated hardware
       • hybrid approach : HAProxy(OSS Load balancer)
         1. Running on client machine(Client managed by HAProxy)
         2. Running on intermediate server(Server Side Components)
         3. HAProxy
            • manages health checks
            • removal & addtion of machines
            • balances requests a/c pools
   • Queues
     • effectively manages requests in large-scal distubuted system
     • In small systems > write are fast
     • In complex systems > high incomming load & individual write take mass time
     • To achieve high performance & availaility (System need to the asynchronous > Queues)
     • difficult for fair & balanced distribution
   • Caching
   • Replication
   • SQL vs No-SQL
     • SQL(Relational Database)
       1. Structured
       2. Predefined schema
       3. Data in rows & columns
       4. MySQL, Oracle, MS SQL Server, SQLite, Postgre, MariaDB
     • NO SQL(Non Relational Dataase)
       1. Unstructured
       2. distributed
       3. dynamic schema
       4. DB Method: Key-Value Stores, Document DB, Wide-column DB, Graph DB
       5. High Level differences between SQL & NoSQL | Property | SQL | NoSQL | |—|—|—| |A|B|C|
   • Indexes
   • Proxies
   • Data Partitioning