Hadoop was built and conceived with well-defined architecture goals and principles, as listed here, (the following are in no way authoritative as we can't find one; rather we gathered this from https://goo.gl/3nvERl):
- Linear scalability (Scale-Out rather than Scale-Up): Add more nodes for scalability to increase data storage and computing power.
- Bring code to data rather than data to code: In big data, data is usually huge and code working on data is small. So, this principle states that bring or distribute code to the nodes/machines where it can act on data and not distribute or move data. In essence, it means minimize data transfer and distribute code instead.
- Deal with failures as they are common: Bring ...