Skip to content

Latest commit

 

History

History
59 lines (54 loc) · 2.74 KB

API_Performance.md

File metadata and controls

59 lines (54 loc) · 2.74 KB
Raw

API Performance

How to improve API performance

  • Optimize database
    • Optimize queries
      • Add indexes to tables.
      • Process denormalization.
    • Optimize connection
      • Implement connection pooling to manage and reuse database connections efficiently, reducing overhead in opening and closing connections for each request.
  • Caching
    • Implement caching mechanisms to store frequently accessed data in memory.
  • Reduce payload size
    • Enable compression (e.g., gzip or Brotli)
    • Use efficient serialization formats like JSON or Protocol Buffers (protobuf).
    • Using pagination, filtering.
  • Load balancing and horizontal scaling
    • Use load balancers to distribute incoming API requests across multiple server instances.
    • Add more API server instances to handle increased traffic. Implement auto-scaling to dynamically adjust server instances based on demand.
  • Rate limiting
    • Implement rate limiting to restrict the number of requests from a single client within a given time frame.
  • Asynchronous processing
    • Offload time-consuming or non-blocking tasks to background jobs or queues, freeing up API resources to handle more requests.
    • Implement asynchronous APIs when appropriate, especially for long-running operations.
  • Content Delivery Networks (CDNs)
    • Use CDNs to distribute and cache static assets like images, videos, and files.
  • Asynchronous logging
    • Synchronous logging deals with the disk for every call and can slow down the system.
    • Asynchronous logging sends logs to a lock-free buffer first and immediately returns. The logs will be flushed to the disk periodically. This significantly reduces the I/O overhead.

Metrics

Response time (Latency)

  • Concepts
    • Response time measures the time it takes for the API to process a request and send back a response.
    • It is often broken down into server processing time and network transfer time.
  • Metrics
    • Average response time
    • 90th or 95th percentile response time
    • Maximum response time

Throughput

  • Concepts
    • Throughput measures the number of requests the API can handle per unit of time (e.g., requests per second or requests per minute). It indicates the API's capacity.
  • Metrics
    • Requests Per Second (RPS)
    • Requests Per Minute (RPM)

Concurrency

  • Concepts
    • Concurrency measures the number of simultaneous requests the API can handle without degrading performance.
  • Metrics
    • Concurrent connections
    • Concurrent users

Data transfer rate

  • Concepts
    • Data transfer rate measures the rate at which data is sent or received by the API, which can impact network performance.
  • Metrics
    • Data sent per second
    • Data received per second