Week 11: Content Delivery

Content Delivery Networks

• Flash crowd problem, slashdot effect
• Network edge
• Caching
• Caching proxy
• Benefits of load balancer
• Multihoming
• Content Delivery Network (CDN)
• Push CDN vs. Pull CDN
• Overlay network
• Mapping system
• Dynamic DNS server
• Akamai domain name lookup
• Origin server
• Transport system
• Parent content servers vs. edge content servers
• Security benefits of a CDN
• Load shedding
• Adaptive bitrate coding (ABR)

Peer-to-Peer Content Delivery

• Gnutella
  • Gnutella basic architecture
  • Content publishing, search, download
  • Forwarding loops
  • Time to Live (TTL)
  • How does Gnutella provide anonymity?
• Kazaa
  • Supernodes vs. ordinary nodes
  • Content publishing, search, download
• BitTorrent
  • Purpose of .torrent file
  • Purpose of seed nodes
  • Purpose of trackers
  • Seeders vs. leechers
  • Content publishing, search, download

Kafka

• Purpose of Kafka
• Publishers (producers), subscribers (consumers)
• Role of message broker
• Topics
• Partitions & partitioned logs
• Durable logs
• Replicated logs: leaders and followers
• Consumers and consumer groups
• Scaling to handle higher message volumes
• Scaling to handle more processing workloads
• Queuing vs. publish-subscribe

Distributed Caching as a Service

• Purpose of a cache
• Memcached - basic role & deployment
  • How a client locates the correct memcached server
  • Eviction and expiration
  • What servers know about each other
  • You don’t need to know the commands
• Redis - basic role & deployment
  • Main features added over memcached
  • You don’t need to know data types but know that they are used
  • Expiration and eviction
  • Transactions
  • Message broker service
  • Range vs. hash partitioning