Exam AWS Certified Solutions Architect - Associate SAA-C03 topic 1 question 507 discussion

Using DynamoDB's global tables feature, you can achieve a globally consistent reservation database with low latency on updates, making it suitable for serving a global user base. The automatic replication provided by DynamoDB eliminates the need for manual synchronization between Regions.

Aurora: less than 1 second: https://aws.amazon.com/rds/aurora/global-database/ DynamoDB: from 0.5 to 2.5 second: https://docs.aws.amazon.com/amazondynamodb/latest/developerguide/V2globaltables_HowItWorks.html

A is the only option that allows you to write from any reagion. I understand the < 1 second latency, but DynamoDB global tables use a last-writer-wins reconciliation between concurrent updates. B and C only supports read scaling, not low-latency global writes. It will be much much much much than 1 second to write on the databse. D I stopped reading at lambda so NO!.

Dynamodb is consistent and read/write in all node, Aurora db , only the master is writable. For reservation , writing is criticle.

A is the correct answer

dynamodb is not globally consistent. DynamoDB doesn't support strongly consistent reads across Regions. https://docs.aws.amazon.com/amazondynamodb/latest/developerguide/V2globaltables_HowItWorks.html

correct answer bbbb

A: dynamoDB global table preview strong consistent https://docs.aws.amazon.com/amazondynamodb/latest/developerguide/multi-region-strong-consistency-gt.html

Aurora replicas have to be in the same region as primary https://docs.aws.amazon.com/prescriptive-guidance/latest/aurora-replication-options/compare-solutions.html

Since DynamoDB Global Tables allow writes in all Regions, conflicts can occur if the same item is updated simultaneously in multiple Regions. DynamoDB uses a "last writer wins" conflict resolution strategy, which may lead to data integrity issues: https://docs.aws.amazon.com/amazondynamodb/latest/developerguide/V2globaltables_HowItWorks.html#V2globaltables_HowItWorks.conflict-resolution If you require less than 1-second latency for write updates across Regions, DynamoDB may not consistently meet this requirement due to the asynchronous nature of its replication. For this scenario, DynamoDB Global Tables do not fully meet the requirements due to their reliance on eventual consistency. If strong consistency is a must, go with Amazon Aurora Global Database.

The question asks "Average latency must be less than 1 second on updates to the reservation database." A is incorrect: " Changes to a DynamoDB global tables are replicated asynchronously, with typical latency of between 0.5 - 2.5 seconds between AWS Regions in the same geographic area." B is the answer: "All Aurora Replicas return the same data for query results with minimal replica lag. This lag is usually much less than 100 milliseconds after the primary instance has written an update."

How can you update your database in the different regions with read replicas? You need to be able to read and write to the database from the different regions.

In my Opinion it is A. The reason is that Aurora Read Replicas support up to 5 Read replicas in different regions . We don't have that limitation with Dynamo DB Global tables, hence I vote for A.

Purely from the wording, seems B. DynamoDB "usually within one second" Aurora "usually less than one second" Question asks for "less than one second" thus Aurora

"a web application for travel ticketing". This would be a transaction, so DynamoDB is not the answer.

Dynamo DB global table acts as a single table. It does not consist of primary and standby databases. It is one single global table which is synchronously updated. Users can write to any of the regional endpoints and the write will be automatically updated across regions. To have a single primary database that is consistent does not align with dynamo db global tables. Option B is even more dumb compared to A since read replicas does not provide failover capability or fast updates from the primary database. The answer almost close to the requirement is Option A even though it is a misfit

The question mentions that the average latency on updates to the regional reservation databases should be less than 1sec. Read replicas provide asynchronous replication and hence the update times will be higher. Hence we can easily scrap all the options containing read replicas from the options. Moreover, a globally consistent database with millisecond latencies screams dynamo db global