Tailored Interview Role Plays Series: Part 5
System Design Scenario #1
Series Part 1 | Series Part 2 | Series Part 3 | Series Part 4
System Design questions are great tool to assess candidate technical know how and expertise. I had been creating role plays designed for specific roles which were actually based on real life experience might be useful for you too:
Let’s assume that you are part of our solution architecture team and a customer without technical background states that he wants to have a solution that can ingest and process 1.5B events per day. He is the CMO at a gaming company and he will use these data for marketing campaigns.
It depends, right? So the candidate needs to start analyzing and investigate the situation.
- Since 1.5B events is a sizable volume, it is important to understand if the customer has enough infrastructure and computing power to handle this. If they do not have, does this data needs to stay on-premise, has sensitive/PII data and has regulatory compliance requirements?
- If the data is not sensitive, no regulation, is it possible to use elastic cloud infrastructure and services?
- Where are the events coming from, is it globally diverse? Any connectivity or technical limitations? Are they coming from similar devices and formats? What are those device tech stack, data formats
- When are the volume peaks, it is peaking at for example at 4am by 1B in 1 hour etc.?
- What is the retention policy, does the data need to stay for processing and reporting for 1 month etc.? Is the customer ok for data storage cost increase for keeping the data longer? When will be data reporting needed, is it realtime, is it near realtime by 5 minutes is ok? Is 1 hour delay while reporting ok? After processing and reporting complete, is it ok to purge the data? Are there any DR expectations? What is the RTO and RPO?
The senior the candidate, the more detailed and investigation questions are expected from them. Assumptions are ok whenever they are stated to the interviewer. The more experience the person has, the more clarification questions, solution pros and cons sharing, business and technical context understanding is expected. Since all projects have a limited budget, cost effectiveness is also important.
This is a quite common streaming analytics and gaming event collection project requirements.
- The reporting delay requirement of the data is critical here.
- Inserting the data in a batch and purging the data as soon as possible could reduce costs.
- Handling scaling requirements is another aspect.
Also check out Pub/Sub: https://cloud.google.com/pubsub#stream-analytics
