Dilbert infamously pointed out, that the internet itself runs in batches. Everything in computers, from TCP to operating systems to databases to the CPUs themselves natively work with batches. Streaming video on the internet gives the illusion of being continuous, but at every layer of the implementation it is processed in batches. They’re just small batches that occur as needed, with latency in seconds not hours or days.
In Business Intelligence we tend to process data in batches. It is the obvious way to do things. It is what everything does natively. I also have experienced the pain of batches:
- Data is always slightly out of date.
- Unattended software has both extra complexity and extra ways in which it can fail.
- And I don’t like getting up in the middle of the night to fix the production warehouse load, particularly when the load is fine but the monitoring system is broken.
Read the rest on the Optimal BI blog.