| I don't agree with the scripter vs developer differentiation. But my interpretation of the difference is that Some developers can read the documentation, put together what some devs might deem easy like build REST API, put on a rate limit, add authorization, integration with payment APIs, scrape website or parse CSV file etc... This is what 80% of the job involves. Some jobs include requirements like this Build a distributed key-value store for storing counters, it must work for 1200 RPS, the counts must be available in both EU and US region (150ms) apart. Counters should be consistent in both the regions. Can you think through such problem? Do you know what it takes?
Do you know updating counter is not a single operation but: 1. Read the original value
2. Increment the value
3. Save the new value Do you know how latency the between the region affects the throughput of the service? How can you make counters reflect consistent count in both regions consistent? What if one of the servers goes down? How do you make it highly available? Customers might increase and we end up needing more nodes. When are more nodes added how is the data rebalanced, what effect does it have on throughput and latency? |
Your interpretation is well said. I am doing that 80% stuff. it seems to deliver a lot and feel like developers currently I am writing very defensive codes which not addressing problem directly. I just want to get it done quickly.