Prepare and test your skills
Prepare and test your skills
Free sample question, answered and explained below. The Professional Cloud Architect practice bank has 2,335 more like it. Try 10 free: instant score, no account.
A global gaming company is architecting a low-latency caching and state store on Google Cloud to handle millions of active sessions. The engineering team requires horizontal write scalability across multiple shards and needs to scale out read throughput across multiple zones while maintaining minimal client response times.
The team decides to deploy Memorystore for Redis Cluster (or Memorystore for Valkey) in Cluster Mode Enabled with read replicas provisioned for each shard, configuring client connections to issue the READONLY command to read directly from replica nodes.
According to distributed systems design trade-offs and the CAP theorem, what architectural trade-off must the team accept with this configuration?
Full cluster freezing is a legacy operational constraint in non-distributed architectures where resharding or capacity resizing requires putting the entire database into read-only maintenance mode and terminating existing client connection pools.
MOVED redirections or topology updates via discovery endpoints.The system were operating on a single-node database that lacks clustering and online resharding capabilities, requiring manual downtime for vertical scaling.
Pessimistic locking is a concurrency control mechanism where transactions acquire explicit database locks on records or keys prior to read or write execution, blocking concurrent operations from modifying or viewing uncommitted changes until the lock is released.
READONLY commands directly to specific replica nodes without cluster-wide lock coordination.The workload were using a document database running in pessimistic concurrency mode where server-side transactions acquire distributed locks across document entities during read operations.
A two-phase commit (2PC) is a distributed consensus protocol used in strongly consistent (CP) relational databases (such as distributed SQL systems) where a coordinator node ensures that all participating nodes synchronously agree to commit a transaction before acknowledging success to the client.
The team chose a distributed strongly consistent database architecture (like Cloud Spanner) where synchronous multi-region or multi-zone consensus is enforced for strict serializability.
In distributed in-memory data stores like Memorystore for Redis Cluster and Memorystore for Valkey, Cluster Mode Enabled partitions data across multiple primary shards (from 1 to 250 shards) to achieve horizontal scalability. To maximize write performance and provide sub-millisecond response times, replication from primary nodes to read replicas is performed asynchronously.
READONLY command, reads do not wait for synchronous replication acknowledgments. If a primary node accepts a write, there is a propagation delay before that write reaches every replica, meaning replica reads may serve stale data.
For ephemeral session states, leaderboards, and caching layers, sacrificing immediate strict consistency in exchange for massive horizontal read/write scale and minimal network latency provides the best performance balance.
Answer them yourself and get your score the moment you finish. No account needed.
Full access includes
Keep the momentum going with these hand-picked practice scenarios
Last updated