Layer 3 · Go deeper · 6 min
Go deeper
Size it, build it, then make the calls an interviewer will push you on.
Build-up · 5 steps
Watch a poller miss four changes
Three polls, ten seconds apart, against four products. The poller sees the table at each poll, not the changes between them.
Step 1 of 5
Three polls
The poller asks for rows with updated_at after its watermark at 0, 10 and 20 seconds. Each poll sees the table as it is at that moment.
4 interview moments
You’re the candidate
The interviewer pushes on the hard parts. Pick an approach and see how it holds up.
Deep dive 1 of 4
Interviewer
“Elasticsearch must track the products table within seconds, deletes included. How do you capture the changes?”
Your move. Pick an approach:
Deep dive 2 of 4
Interviewer
“A new connector must first copy a 400-million-row table that keeps changing. How do you snapshot it?”
Your move. Pick an approach:
Deep dive 3 of 4
Interviewer
“You are about to create the first replication slot on your PostgreSQL primary. What do you set up first?”
Your move. Pick an approach:
Deep dive 4 of 4
Interviewer
“The payments team wants to react when an order is placed. Do you point them at your CDC topics?”
Your move. Pick an approach:
Build-up · 4 steps
One chunk of an incremental snapshot
Netflix’s DBLog idea, adopted by Debezium. The chunk is read between two watermarks while log events keep flowing.
Step 1 of 4
Open the window
Debezium writes a low watermark into the log. Log events are never held back; u p1 goes straight to the topic.