Determine high-performing database solutions

Hollins University runs a course-registration system on a self-managed MySQL server. Load is almost zero for most of the year, then jumps by two orders of magnitude for the two-week registration windows each semester. The application relies on multi-table joins and transactions. The university wants a managed, MySQL-compatible database that scales capacity up and down automatically in seconds and charges for the capacity actually used.

Which solution will meet these requirements MOST cost-effectively?

1. AAmazon RDS for MySQL on the largest available instance classA fixed large instance is paid for all year even though it is idle most of the time, and resizing it requires a restart rather than seconds-scale automatic scaling.
2. BAmazon DynamoDB in on-demand capacity modeOn-demand DynamoDB scales and bills on use, but the application depends on joins and multi-table transactions that would require a complete redesign away from the relational model.
3. CAmazon Aurora Serverless v2 for Aurora MySQLAurora Serverless v2 is MySQL-compatible, scales capacity in fine-grained Aurora capacity units within seconds based on load, and bills per second for the capacity consumed, which fits a workload that is idle most of the year.
4. DAmazon Aurora MySQL provisioned instances covered by a three-year Reserved InstanceReserved pricing lowers the hourly rate of a fixed-size instance, but the university would still pay for peak-sized capacity during the long idle periods.

Dunmere Logistics is building a fleet-routing application that stores vehicle positions and delivery zones and needs to run geospatial queries such as finding every vehicle inside a polygon using the PostGIS extension. The application also stores semi-structured shipment manifests that it queries with JSONB operators. Dunmere wants a managed relational database with high throughput, automatic storage scaling, and the fewest changes to the code its developers already wrote.

Which database engine should a solutions architect recommend?

1. AAmazon Aurora PostgreSQL-Compatible EditionAurora PostgreSQL supports the PostGIS extension and JSONB natively, so the existing queries run unchanged, and Aurora adds high throughput, automatic storage growth, and up to 15 replicas.
2. BAmazon Aurora MySQL-Compatible EditionMySQL has its own spatial and JSON functions, but PostGIS and JSONB are PostgreSQL-specific, so the queries would have to be rewritten.
3. CAmazon RDS for MariaDBMariaDB is a MySQL-family engine without PostGIS or JSONB, and it lacks the Aurora storage architecture the throughput requirement points to.
4. DAmazon DynamoDB with a geohash sort keyDynamoDB can support location lookups through geohashing libraries, but it is not relational, has no polygon queries, and would require a complete rewrite.

Winterbourne Retail stores its product catalog in an Amazon DynamoDB table named catalog-items. The storefront performs millions of eventually consistent reads per hour against a small set of popular items, and the company wants to reduce read latency from milliseconds to microseconds and cut DynamoDB read costs. The development team wants to change as little application code as possible.

What should a solutions architect recommend?

1. ADeploy an Amazon ElastiCache for Valkey cluster and add cache-aside logic to the applicationElastiCache can cache the items, but the application must be modified to check the cache, handle misses, and invalidate entries, which is more code change than the team wants.
2. BIncrease the table's provisioned read capacity unitsMore read capacity raises throughput at higher cost but does not reduce the per-request latency of DynamoDB itself.
3. CEnable DynamoDB global tables and read from the nearest replicaGlobal tables reduce cross-Region latency; they do not provide microsecond in-memory reads and would increase cost through replicated writes.
4. DCreate a DynamoDB Accelerator (DAX) cluster and point the application at the DAX endpoint using the DAX clientDAX is a write-through, API-compatible in-memory cache for DynamoDB; swapping the SDK client for the DAX client delivers microsecond reads for repeated items and reduces reads against the table with minimal code change.

Saltmarsh Media is launching a new mobile app whose backend stores user activity in an Amazon DynamoDB table named activity-events. The company has no traffic history, expects sudden spikes when the app is featured in app stores, and wants requests to succeed during those spikes without throttling. Saltmarsh also wants to pay only for the requests it actually makes rather than for pre-planned capacity.

How should a solutions architect configure the table?

1. AUse on-demand capacity modeOn-demand mode accommodates workloads with unknown or spiky traffic, instantly serves up to double the previous peak, and bills per read and write request with no capacity planning.
2. BUse provisioned capacity mode with DynamoDB auto scaling between 5 and 1,000 capacity unitsAuto scaling adjusts provisioned throughput in response to CloudWatch alarms over several minutes, so an abrupt spike can be throttled before the scale-up completes, and Saltmarsh would pay for provisioned units even when idle.
3. CUse provisioned capacity mode at the maximum expected peak throughputProvisioning for peak guarantees capacity but pays for it continuously, which contradicts the pay-per-request requirement and is impossible to size without traffic history.
4. DCreate a global table replica in a second Region to absorb the spikesGlobal tables provide multi-Region replication and local access; they do not change how capacity is allocated or billed within a Region.