Multi-Service Tradeoff — Azure Solutions Architect (AZ-305)
ACI, AKS, and Functions Across Workload Lifecycle Shapes
Short-lived isolated tasks with no inter-service communication fit Azure Container Instances: no cluster to provision or maintain, billed per second of execution. AKS is correct when the workload requires service mesh, horizontal pod autoscaling, or persistent inter-service routing that outlasts individual requests. Azure Functions removes infrastructure ownership entirely for stateless, event-driven handlers with execution times under 10 minutes. Queue Storage decouples producers from consumers but is not a compute service. Map workload lifecycle and operational maturity to the compute model before comparing features. The scenario description of lifecycle and team maturity does that mapping for you.
What This Pattern Tests
Azure offers three messaging services with distinct models. Service Bus handles enterprise messaging with sessions, dead-lettering, and exactly-once delivery at $0.05 per million operations. Event Grid handles reactive event routing with push delivery at $0.60 per million events. Queue Storage handles simple FIFO queueing at $0.004 per 10,000 transactions. The exam gives you a messaging requirement and tests whether you match it: "order processing with dead-letter handling" = Service Bus, "react to blob uploads" = Event Grid, "simple task queue for background workers" = Queue Storage. Cosmos DB vs. SQL Database vs. Table Storage follows the same principle: global multi-model vs. relational with joins vs. simple key-value.
Decision Axis
Message complexity and delivery model determine service. Over-specifying is as wrong as under-specifying.
Associated Traps
Decision Rules
Determine whether Azure SQL Database General Purpose with zone-redundant configuration already satisfies the stated 99.99% SLA, making Business Critical tier or Azure SQL Managed Instance unnecessary and cost-violating over-provisioning.
Whether the batch ingestion workload scale and team skill profile justify a self-managed Spark cluster (Databricks) or whether a managed low-code data flow service (ADF mapping data flows) satisfies all stated constraints at lower operational cost.
Which Azure SQL Database tier and zone-redundancy combination satisfies RPO less than 10 minutes and RTO less than 30 minutes at minimum monthly cost, without upgrading to a higher-priced tier whose additional HA properties are unnecessary given the stated numeric targets?
Whether a managed low-code orchestration service (Azure Data Factory mapping data flows) or a self-managed Spark platform (Azure Databricks) better satisfies a batch workload of moderate scale when the team lacks cluster-management expertise and the scenario forbids operational overhead.
Whether the cross-region RPO ≤ 5 min target is satisfied by a tier upgrade to Business Critical with zone redundancy (intra-region HA only) or by Active Geo-Replication on a General Purpose tier (cross-region near-zero RPO at lower compute cost).
Does an RPO ≤ 5 minutes for a data integration pipeline mandate a continuous streaming architecture, or can a managed incremental CDC pipeline running on sub-5-minute tumbling windows satisfy the constraint within the team's operational capability?
Whether Azure SQL Database General Purpose with zone-redundant deployment satisfies the stated 99.99% availability SLA and RPO ≤ 15 minutes for zone failures—anchored on the fact that GP zone-redundant already achieves the 99.99% SLA and that automated differential backups running every 5–12 minutes satisfy RPO ≤ 15 min, making Business Critical an unjustified cost premium that violates the cost-minimization directive.
Whether pipeline run frequency alone satisfies the stated cross-region RPO, or whether ADLS geo-redundant replication tier plus a secondary-region ADF pipeline recovery configuration must be explicitly provisioned to guarantee data currency and pipeline resumption after a regional outage.
Whether to deploy Azure SQL Database on the Business Critical tier (synchronous zone-redundant local SSD replicas, RPO=0) or the General Purpose tier (remote asynchronous storage replication, RPO>0) paired with Zone-Redundant Blob Storage, given an explicit RPO=0 constraint on the relational tier.
Whether to use synchronous zone-redundant replication at both the relational tier (Business Critical + zone-redundant config) and the blob tier (ZRS), versus selecting near-right options that satisfy availability SLA on paper but use asynchronous replication and therefore cannot guarantee RPO=0.
Whether to deploy Azure SQL Database at the Business Critical tier with zone redundancy enabled (synchronous Always On AG replication across zones, RPO=0) versus General Purpose tier with auto-failover groups (async geo-replication, RPO measured in minutes), combined with whether to use Azure Blob Storage ZRS (synchronous zone replication, RPO=0) versus LRS or GRS (async or single-datacenter, RPO > 0).
Whether the workload's combination of container packaging, event-driven burst scaling, scale-to-zero economics, and a team with zero Kubernetes operational capacity disqualifies AKS and makes Azure Container Apps the only tier that satisfies all stated constraints simultaneously.
Whether Azure Batch's managed node-pool lifecycle, native task scheduling, and auto-scale-to-zero capability satisfies both the workload-density-target and the operational-complexity-budget constraint better than AKS with a batch job controller or self-managed VMs.
Whether Azure Container Apps or AKS is the correct compute tier for a containerized, scale-to-zero, stateless microservice when the team explicitly lacks Kubernetes cluster management expertise and cannot absorb ongoing cluster operational burden.
Whether the chosen compute tier and redundancy model satisfies the explicit cross-region RPO and RTO targets rather than the within-region zone-level HA guarantee that a zone-redundant App Service plan provides.
Domain Coverage
Difficulty Breakdown