Exercises › Multi-Region Deployment Language › Multi-Region Data Vocabulary

Multi-Region Data Vocabulary

5 questions · Multi-Region Deployment Language

1. A database architect says: "We chose eventual consistency for our user profile service across regions." What does this mean for reads?

Reads will always return the most up-to-date value immediately after a write After a write, reads from different regions may temporarily return stale data — but all replicas will converge to the same value over time Writes are queued and applied to all regions at the same time to ensure readers always see the same data

2. An engineer explains: "We use read replicas in eu-west-1 and ap-southeast-1, with the primary in us-east-1." What can read replicas do that the primary cannot be replaced by?

Read replicas can accept writes from all regions simultaneously without conflict Read replicas serve read queries from nearby regions with lower latency, but all writes must go to the primary Read replicas provide strongly consistent reads that are guaranteed to be more up-to-date than the primary

3. A team implements DynamoDB Global Tables. What is the key capability this provides?

A single DynamoDB table that is accessible from all AWS regions via a global endpoint, with all data stored in us-east-1 Multi-master replication where each region has a full copy of the table and can accept writes — DynamoDB handles conflict resolution automatically A read-only replica of a primary table, distributed across regions for low-latency reads

4. Two engineers debate conflict resolution in multi-master replication. The first says "last write wins"; the second argues for CRDTs. What problem are they solving?

How to merge network packets from different regions that arrive out of order What to do when two regions independently update the same piece of data before they have synchronised — which version wins, and can updates be merged automatically? How to split data between regions so that each region only stores the data relevant to its users

5. A senior engineer explains: "For our financial transactions, we need strong consistency across regions." Why is this harder to achieve than eventual consistency?

Strong consistency requires more storage space per region because each region stores multiple versions of each record Strong consistency requires a write to be acknowledged by all (or a quorum of) replica regions before succeeding — adding cross-region network latency to every write operation Strong consistency can only be achieved with a single-region database — it is architecturally impossible in multi-region setups

Exercise complete!

out of 5 questions

← Back to Multi-Region Deployment Language All exercises