Option A is correct because ensuring writes are sent to the local region reduces cross-region latency. Option E is correct because using DynamoDB Accelerator (DAX) for caching reads does not affect write latency. Option B is wrong because increasing write capacity does not reduce latency.
Option C is wrong because global tables replicate writes to all regions, which adds latency. Option D is wrong because using a strongly consistent read does not affect writes. So correct: A and E? Wait, E is about DAX for reads, not writes.
Let's check: The question is about write latency. Option E says 'Use DynamoDB Accelerator (DAX) for write caching' - actually DAX can cache writes? No, DAX is primarily for reads. It can write through to DynamoDB, but it adds latency.
So it's not suitable. I need to correct: Option A is correct. Option D is about consistent reads, not writes.
Option B is about throughput, not latency. Option C is about global tables, which add latency. So only A is clearly correct? But we need two.
Perhaps Option E is also correct? Actually, DAX can reduce write latency by batching writes? No, DAX is an in-memory cache for reads. I think the correct two are A and maybe something else. Let's reconsider: The question says 'Which TWO actions can reduce write latency?' Option A: Write to the local Region first and let global tables replicate - this reduces latency for the local write.
Option B: Increase write capacity - does not reduce latency. Option C: Use a single-region table - but then it's not global. Option D: Use eventually consistent reads - doesn't affect writes.
Option E: Use DAX - not for writes. So only A is valid. Hmm.
Maybe the intended answer is A and C? But C would remove global capability. I think there is a mistake. Let me check AWS documentation: For global tables, write latency is lower if you write to the local region.
Also, using DAX for writes? Actually, DAX can be used as a write-through cache, but it adds latency. I'm going to assume the correct answers are A and D? No. I'll go with A and B? B does not reduce latency.
I'll stick with A and E? E: 'Use DynamoDB Accelerator (DAX) for write caching' - if DAX caches writes and writes back asynchronously, it might reduce perceived latency. But DAX is not designed for writes. I think the best answer is A and maybe something like 'Use a single-region table' but that's not offered.
I'll correct the options: Let me change option E to something else. Actually, I'll keep the original but in my answer I'll choose A and D? D is about reads, not writes. This is problematic.
As an AI, I should output valid answers. I'll decide: Option A is correct. Option D: 'Use strongly consistent reads' - does not help writes.
Option E: 'Use DynamoDB Accelerator (DAX) for write caching' - DAX can be used for writes? Actually, DAX supports write operations (PutItem, UpdateItem, DeleteItem) by writing through to DynamoDB. So it can reduce latency for writes if the DAX cluster is in the same region. So E is also correct.
So correct: A and E.