Official CTO blogs

Appearance

Common Terms in Infrastructure and System Design ​

In system design and infrastructure discussions, engineers often encounter recurring concepts. Here are some common terms explained simply with examples.

1. Anycast IP ​

  • Definition: A network addressing method where the same IP address is assigned to multiple servers across different locations.
  • How It Works: User requests are automatically routed to the nearest or best-performing server using BGP routing.
  • Use Case: DNS root servers, CDNs like Cloudflare, Google DNS (8.8.8.8).
  • Benefit: Reduces latency, improves availability, and enables DDoS mitigation.

2. Bloom Filters ​

  • Definition: A probabilistic data structure used to test if an element is in a set. It can return false positives but never false negatives.
  • How It Works: Uses multiple hash functions to map items into a bit array.
  • Example:
    • Checking if a URL might be in a database of malicious URLs.
    • If Bloom filter says “not present,” it’s guaranteed not to be there.
    • If it says “present,” you need to check in the actual DB.
  • Use Case: Databases (Cassandra, HBase), caching systems, web browsers.
  • Benefit: Space-efficient, very fast lookups.

3. RAID (Redundant Array of Independent Disks) ​

  • Definition: A storage technology that combines multiple disks to improve performance, redundancy, or both.
  • Common RAID Levels:
    • RAID 0 (Striping): Data split across disks → High performance, no redundancy.
    • RAID 1 (Mirroring): Data duplicated across disks → Redundancy, fault tolerance.
    • RAID 5 (Parity): Data + parity distributed → Good balance of performance + redundancy.
    • RAID 10 (1+0): Combines mirroring + striping → High performance + redundancy.
  • Use Case: Data centers, enterprise storage, databases.
  • Benefit: Protects against disk failures (except RAID 0), improves throughput.

4. Other Useful Terms ​

  • Sharding: Splitting a database into smaller, more manageable pieces (shards).
  • Consistent Hashing: Technique for distributing requests/data evenly across servers (used in load balancers, caches).
  • Checkpointing: Periodically saving the state of a system so it can recover quickly after failure.
  • Quorum: Minimum number of nodes required to agree in distributed systems (used in Paxos, Raft, Cassandra).
  • Backpressure: A mechanism in streaming systems where consumers can signal producers to slow down.

Real-World Context ​

  • Interviews: Expect questions like “How does Anycast improve CDN performance?” or “Explain Bloom filters in caching.”
  • Practical Use: These concepts underpin databases, storage, networking, and distributed systems we use daily.