Unlocking the Value of Real-Time Analytics

Chapter 2. Typical Trade-offs and Associated Workarounds

Now that we understand the requirements of real-time analytics (latency, freshness, throughput, and concurrency), we can dive deeper to determine where delays between data collection and actionable analytics may occur. We can examine common traditional steps companies take to mitigate these problems. Remember, our goal is to keep the four requirements of real-time analytics as functional and effective as possible. Delays between data collection and decisions from it mean lost time, less money, and missed opportunities.

Freshness for Latency

Speed was the first of our requirements for successful real-time analytics. We mentioned how large datasets combined with complex and multiple-table queries can slow down data pulls from your database. How do companies avoid encountering these issues? Your analysts and application developers understand what data is needed, and they pull everything together with their queries. If that’s so, how can you make their queries faster and more efficient without cutting valuable information from the results?

Views

One method companies employ is designing a view. In short, the developers will take requirements from business analysts regarding frequently used and long-running queries. From these requirements, they can create views. Views are simply queries against the data that can be stored as tables or run on command on the server. Views drastically improve query response times while decreasing ...