How to Meet a 10-Second Bedrock Batch Analysis SLA?

AIP-C01 By islever Editorial Last updated Oct 18, 2026

Implement application integration patterns and development tools. Optimize application performance.

Short answer

B Use a Step Functions Parallel state to invoke Bedrock analysis Lambda functions concurrently, with Bedrock timeouts and CloudWatch latency metrics.

Question

A financial services company needs to build a document analysis system that uses Amazon Bedrock to process quarterly reports. The system must analyze financial data, perform sentiment analysis, and validate compliance across batches of reports. Each batch contains 5 reports. Each report requires multiple foundation model (FM) calls. The solution must finish the analysis within 10 seconds for each batch. Current sequential processing takes 45 seconds for each batch.
Which solution will meet these requirements?

Use AWS Lambda functions with provisioned concurrency to process each analysis type sequentially. Configure the Lambda function timeouts to 10 seconds. Configure automatic retries with exponential backoff.
Use AWS Step Functions with a Parallel state to invoke separate AWS Lambda functions for each analysis type simultaneously. Configure Amazon Bedrock client timeouts. Use Amazon CloudWatch metrics to track execution time and model inference latency. Correct Answer
Create an Amazon SQS queue to buffer analysis requests. Deploy multiple AWS Lambda functions with reserved concurrency. Configure each Lambda function to process different aspects of each report sequentially and then combine the results.
Deploy an Amazon ECS cluster that runs containers that process each report sequentially. Use a load balancer to distribute batch workloads. Configure an auto-scaling policy based on CPU utilization to handle demand fluctuations.

Key Takeaways

The test is spotting that a 4.5x latency gap requires explicit parallel fan-out across analysis tasks, not warmed Lambda capacity or more queue consumers; options that keep the per-report chain sequential are traps.

This question asks how to orchestrate Amazon Bedrock document analysis so a five-report batch finishes in 10 seconds instead of the current 45 seconds. The page establishes that a Step Functions Parallel state with concurrent Lambda branches is the correct answer (B) over sequential Lambda, SQS buffering, or ECS containers.

Common mistake: The common wrong pick is A because provisioned concurrency sounds like a performance fix, but it only reduces cold starts; the financial, sentiment, and compliance calls still execute sequentially and a 10-second Lambda timeout can cut them off.

Expert Analysis

Why the Answer Is Correct

Option B is the only choice that explicitly removes the sequential bottleneck: a Step Functions Parallel state fans out separate Lambda functions for financial analysis, sentiment analysis, and compliance validation so their Bedrock FM calls overlap instead of running one after another. It also keeps the operational controls the scenario needs, including Bedrock client timeouts to cap slow model calls and CloudWatch metrics to confirm the batch stays under the 10-second target. Within this option set, that is the intended horizontal-scaling pattern for a 45-second to sub-10-second requirement, even if a Step Functions Map state over the five reports could provide additional report-level fan-out.

Why the Other Options Are Wrong

Option A keeps the analysis types sequential, so provisioned concurrency only reduces cold starts and does not fix the 45-second critical path; a 10-second Lambda timeout can even truncate work. Option C adds an SQS buffer and reserved concurrency, but it still describes processing aspects of each report sequentially and adds queue latency without an explicit parallel orchestration step. Option D runs containers that process each report sequentially and relies on CPU-based auto-scaling plus a load balancer, which does not collapse the 45-second chain within a strict 10-second SLA. None of A, C, or D introduces the explicit Parallel state that makes the Bedrock analysis branches run concurrently.

Community Comment Notes

Several learners converged on B: eesa described the Parallel state as the best solution because it processes financial, sentiment, and compliance analyses simultaneously instead of sequentially, and tumi98 emphasized concurrent execution plus Bedrock timeouts and CloudWatch. Chibuzo1 framed the requirement as roughly a 4.5x speedup that only parallelization can deliver. blieskies raised a worthwhile nuance, saying "I wish they'd suggested MAP instead of PARALLEL," which is a reminder that Map over the five reports may be an even tighter design, but Parallel remains the best fit among the offered options. GiorgioGss dismissed the alternatives succinctly because they add latency.

Official Reference

These links open the official vendor documentation in a new tab.

Exam Strategy

When a question gives a current time and a target time, calculate the required speedup before reading the options; if it is greater than 1, look for explicit parallel branches or Map/Parallel states. Treat provisioned and reserved concurrency as throughput controls, not as latency reducers for a single sequential chain.

Frequently Asked Questions

Why doesn't provisioned concurrency in Option A meet the 10-second batch requirement?

Provisioned concurrency keeps Lambda instances warm but still runs financial, sentiment, and compliance analysis sequentially; it reduces cold starts, not the 45-second critical path.

Would an SQS queue with reserved concurrency in Option C be enough to parallelize five reports?

It can increase throughput, but it does not by itself orchestrate the three analysis types in parallel, and queue buffering adds latency to a strict 10-second per-batch SLA.

More AIP-C01 FAQ →

Discussion 9 contributions

Compiled from public community threads and reader submissions, lightly edited for clarity. Handles and vote tallies are shown as recorded, and comments reflect the views of individual contributors.

From public community threads 9

eesa 👍 2 Answered B

AWS Step Functions con Parallel state (Opción B) - MEJOR SOLUCIÓN: ✅ Paralelización efectiva: Procesa múltiples análisis (financiero, sentimiento, compliance) simultáneamente en lugar de secuencialmente ✅ Reduce tiempo drásticamente: De 45s a ~10s al ejecutar tareas en paralelo ✅ Orquestación robusta: Step Functions maneja la coordinación de múltiples Lambda functions ✅ Monitoreo integrado: CloudWatch metrics para tracking de latencia y tiempos ✅ Manejo de errores: Retry logic y error handling nativos ✅ Escalabilidad: Cada tipo de análisis escala independientemente
a9fb7b2 👍 2 Answered B

Step Functions workflow can invoke separate Lambda functions in parallel for different analysis tasks, which is exactly what this batch pattern needs. Can also monitor execution time and Bedrock/model latency with CloudWatch metrics, which helps verify the system stays under the 10-second target. Amazon Bedrock exposes monitoring through CloudWatch, and CloudWatch generative AI observability includes latency views for model invocations.
GiorgioGss 👍 2 Answered B

ACD fail because it ads latency.
Chibuzo1 👍 1 Answered B

The core problem is a 45-second sequential process that must finish in 10 seconds — a 4.5x speedup requirement. The only way to achieve this without faster hardware is parallelization. Option B directly addresses this with Step Functions' Parallel state, which fans out multiple Lambda invocations simultaneously across analysis types (financial data, sentiment, compliance). Since these three workloads are independent of each other, running them concurrently collapses wall-clock time dramatically. CloudWatch integration for latency tracking is a practical operational addition, not overhead.
blieskies 👍 1 Answered B

I accept it's B because this processing required parallelisation but I wish they'd suggested MAP instead of PARALLEL because it we assume that each step for a report takes the same amount of time then it'd still take 45/3 seconds = 15 seconds .. whereas parallel would have given us a tidy 45/5 = 9 seconds (under the SLA) requirement

Show 4 more comments

tumi98 👍 1 Answered B

Independent analysis tasks execute concurrently, reducing total batch processing time. Bedrock timeouts and CloudWatch provide appropriate operational controls.
de1612d 👍 1 Answered B

I want to parallelize batch processing, so the answer is B.
67bdb19 👍 1 Answered C

Feels like C is the one. Does anyone disagree?
Neos03 👍 1 Answered A

A is correct

Reader comments & corrections

No reader corrections yet — spotted an error or have a note? Share it below.

Practice All AIP-C01 Questions

Access 110 questions with complete answers and detailed explanations.

View Full AIP-C01 Practice Test →

← Back to AIP-C01 Study Guide