Scale up compute resources by increasing vCores is the correct option.

Scale up compute resources by increasing vCores reduces write latency because adding vCores increases CPU and memory and it usually raises the database service limits for log throughput and IOPS in the vCore purchasing model. Increasing compute addresses transactional and log flush bottlenecks that cause elevated write latency during peak sale events and it is the standard way to improve write performance for an Azure SQL Database.

Increase the database storage allocation is incorrect because adding storage capacity gives more space but does not necessarily improve write performance. Storage allocation increases disk size but Azure SQL Database IOPS and log throughput are tied to service tier and compute, so simply growing storage will not resolve compute or log related write latency.

Provision read-only replicas for load distribution is incorrect because read replicas spread read workload and do not accept write traffic. Read scaling helps read latency and throughput but it will not reduce write latency on the primary database during heavy transactional peaks.

Increase storage throughput settings is incorrect for a single Azure SQL Database because there is no independent per-database storage throughput knob in most service tiers. Storage performance is typically governed by the compute tier, service tier, or specific architectures like Hyperscale, so the practical way to lower write latency is to scale compute rather than attempting to adjust a separate storage throughput setting.

Full AWS Practitioner Certification Question