Consider the following hypothetical scenario: A Web application receives 25% write traffic and 75% read traffic. Given that this application can tolerate reads that slightly stale, an easy way to scale the database throughput is to introduce slaves and use them for the read-heavy traffic. Now what happens if, say the application traffic pattern is such that it receives 25% writes, 50% reads that must be up-to-date, and 25% reads that can be stale. How would you go about scaling this?
Brian
scalability
