Troubleshooting and Optimization — DVA-C02 Practice Quiz

Question 1

A developer maintains a REST API on Amazon API Gateway that serves a product catalog. The GET /products endpoint receives thousands of identical requests per minute, and each request invokes a Lambda function that reads from a database. The catalog data changes only a few times per day. The team wants to reduce Lambda invocations, lower latency, and cut cost with minimal code changes. What is the MOST appropriate solution?

Accepted Answer

Enable API Gateway caching on the stage for the GET /products method with an appropriate TTL and cache key settings.

Answer

Configure Lambda provisioned concurrency to keep instances warm for the GET /products integration.

Answer

Add an ElastiCache for Redis cluster inside the Lambda function and implement a cache-aside pattern in the function code.

Answer

Increase the Lambda function's memory allocation to speed up each database read.

Question 2

A developer runs a media website on an Application Load Balancer backed by EC2 instances. Product images and CSS/JS files rarely change but are requested millions of times daily by users across many countries. Origin CPU is saturated serving these static files, and users far from the region report slow page loads. Which change most effectively reduces origin load and improves global latency for these assets?

Accepted Answer

Place an Amazon CloudFront distribution in front of the ALB and set appropriate cache behaviors and TTLs for the static assets

Answer

Add an Amazon ElastiCache for Redis cluster and have the EC2 instances read the images from the cache before serving them

Answer

Enable API Gateway caching in front of the ALB to cache the static file responses

Answer

Provision Amazon DynamoDB Accelerator (DAX) to cache the image metadata and reduce database round trips

Question 3

A developer supports a Lambda function that processes uploaded orders. During flash sales, customers report failed order submissions. In CloudWatch, the developer observes that the function's Throttles metric spikes to hundreds per minute while the Errors metric remains near zero, and the ConcurrentExecutions metric flatlines at a fixed value during the spikes. What is the most likely root cause of the failed submissions?

Accepted Answer

The function is hitting its concurrency limit and being throttled, so invocations are rejected before the handler runs.

Answer

The function code is throwing unhandled exceptions that cause each invocation to fail after processing begins.

Answer

The function is experiencing cold starts that cause each invocation to time out during initialization.

Answer

The function's IAM execution role is missing permissions, causing every invocation to be denied.

Question 4

A developer maintains a REST API backed by Lambda and DynamoDB. Support tickets report that the API feels slow for some users, but the CloudWatch alarm monitoring the Average of the API Gateway Latency metric never triggers because the mean stays low. The team wants an alarm that reliably catches the intermittent slowness affecting a minority of requests. What is the most effective change?

Accepted Answer

Create the alarm on the p99 percentile statistic of the Latency metric instead of the Average.

Answer

Increase the alarm's evaluation period so the Average is calculated over a longer window.

Answer

Switch the alarm to use the Sum statistic of the Latency metric.

Answer

Lower the alarm threshold on the Average statistic to a smaller millisecond value.

Question 5

A developer needs to publish custom application metrics (such as order value and items-per-order) from a high-throughput Lambda function. The function already writes structured JSON logs to CloudWatch Logs. Calling the PutMetricData API on every invocation has caused throttling and added latency to the function. What is the MOST efficient way to emit these custom metrics while avoiding the API throttling and extra latency?

Accepted Answer

Write log entries in the CloudWatch embedded metric format (EMF) so CloudWatch automatically extracts the values as custom metrics from the logs.

Answer

Batch the metrics in memory and call PutMetricData once per minute using a background thread inside the Lambda function.

Answer

Enable Lambda enhanced monitoring so the order value and items-per-order are captured as built-in metrics.

Answer

Create a CloudWatch metric filter on the log group that increments a metric each time the function logs an order event.

Question 6

A developer runs a Lambda function triggered by API Gateway. Users intermittently receive 502 errors during traffic peaks. The developer suspects the function is failing to keep up with demand. Which CloudWatch metric should the developer examine first to confirm whether requests are being rejected before they can execute?

Accepted Answer

The Throttles metric for the Lambda function

Answer

The Duration metric for the Lambda function

Answer

The IntegrationLatency metric for the API Gateway stage

Answer

The Errors metric for the Lambda function

Question 7

A developer is investigating an increase in HTTP 500 errors reported by an application running on AWS Lambda behind API Gateway. The Lambda function writes structured JSON logs (including a 'level' field and 'errorType' field) to Amazon CloudWatch Logs. The developer needs to quickly identify which specific error type is most frequently associated with the failures over the last hour, without exporting logs to another service. What is the FASTEST way to do this?

Accepted Answer

Run a CloudWatch Logs Insights query against the function's log group that filters where level = 'ERROR', then uses stats count(*) by errorType and sorts by the count in descending order

Answer

Create a CloudWatch metric filter on the log group and wait for a new custom metric to accumulate data points before graphing it

Answer

Enable AWS X-Ray active tracing and inspect the service map to count error types across recent traces

Answer

Stream the log group to Amazon S3 with a subscription filter and query the objects using Amazon Athena

Question 8

A developer runs a Java application on Amazon EC2 that writes plain-text logs to a CloudWatch Logs log group. Operations wants to be paged whenever the phrase "OutOfMemoryError" appears more than 3 times within a 5-minute window. There is currently no numeric metric that tracks these occurrences. What is the most appropriate way to enable this alerting with the least custom code?

Accepted Answer

Create a CloudWatch Logs metric filter that matches "OutOfMemoryError" and increments a custom metric, then create a CloudWatch alarm on that metric with a threshold of 3 over a 5-minute period.

Answer

Enable CloudWatch Logs Insights scheduled queries to count occurrences and configure the query to publish an SNS notification directly when the count exceeds 3.

Answer

Add the CloudWatch agent to the instance and configure it to emit an alarm whenever it detects the "OutOfMemoryError" string in the log stream.

Answer

Create an EventBridge rule that pattern-matches the "OutOfMemoryError" log events delivered from CloudWatch Logs and triggers an SNS topic after 3 matches.

Question 9

A developer's Lambda function currently writes plain-text log lines such as 'Processed order 12345 in 240ms'. The operations team wants to build CloudWatch Logs Insights queries and metric filters that can reliably extract the order ID, duration, and status without brittle string parsing. What is the MOST effective change the developer should make to the logging approach?

Accepted Answer

Emit each log entry as a single-line JSON object with named fields (e.g. orderId, durationMs, status) so downstream tools can parse and query discrete attributes.

Answer

Increase the Lambda memory allocation so logs are flushed to CloudWatch more quickly and become easier to parse.

Answer

Write logs across multiple lines per event with one attribute per line to make each value visually distinct.

Answer

Enable X-Ray active tracing so the log lines are automatically converted into structured segments.

Question 10

A developer manages a Lambda-based order processing service whose traffic varies predictably across the day, with higher volume during business hours and much lower volume overnight. The team wants a CloudWatch alarm on the function's error count that adapts to these normal traffic patterns and only alerts when errors deviate meaningfully from expected behavior, without constant false alarms from a fixed threshold. What should the developer configure?

Accepted Answer

Create a CloudWatch alarm using anomaly detection on the Errors metric with a band based on the model, alarming when values fall outside the band.

Answer

Create a static-threshold CloudWatch alarm on the Errors metric set to a high fixed value to avoid daytime noise.

Answer

Create a CloudWatch composite alarm combining the Invocations and Duration metrics with fixed thresholds.

Answer

Enable Lambda enhanced monitoring and rely on the default dashboard graphs for manual review each morning.

Question 11

A developer maintains a distributed application composed of an API Gateway REST API, a front-end Lambda function, an SQS queue, and a downstream processing Lambda function. Customers occasionally report failed requests, but the team cannot trace an individual customer request across all four components because each service logs independently to its own CloudWatch Logs group. What is the MOST effective way to enable end-to-end tracing of a single request across these services?

Accepted Answer

Generate a unique correlation ID at the API entry point, propagate it through the SQS message attributes and Lambda payloads, and include it in every structured log statement.

Answer

Enable CloudWatch Logs Insights and query all four log groups simultaneously using the request timestamp to correlate entries.

Answer

Increase the CloudWatch Logs retention period on all four log groups so older entries remain available for manual correlation.

Answer

Configure a single shared CloudWatch Logs group and have all four services write their logs to it.

Question 12

A developer built a reporting feature that repeatedly retrieves all orders for a specific customer from a DynamoDB table. The table's partition key is OrderId and it has no secondary indexes. The feature currently uses a Scan operation with a FilterExpression on CustomerId, and users report that the report takes several seconds and consumes a large amount of read capacity as the table grows. What is the most effective way to reduce both latency and read consumption?

Accepted Answer

Create a global secondary index with CustomerId as the partition key and query that index instead of scanning the table.

Answer

Switch the Scan to use eventually consistent reads and increase the page size to retrieve more items per request.

Answer

Enable DynamoDB Accelerator (DAX) so the Scan results are cached and served with microsecond latency.

Answer

Increase the table's provisioned read capacity units so the Scan completes faster under load.

Question 13

A developer maintains a product catalog service backed by DynamoDB with provisioned capacity. Read traffic has grown, and the table is now consuming most of its provisioned read capacity units (RCUs), driving up cost. Analysis shows the application always calls GetItem with ConsistentRead set to true, but the catalog data changes only a few times per day and stale reads of a second or two are acceptable. What change will most cost-effectively reduce read capacity consumption?

Accepted Answer

Switch the reads to eventually consistent reads by removing ConsistentRead=true, so each read consumes half an RCU per 4 KB.

Answer

Enable DynamoDB Streams on the table to offload read traffic to a downstream consumer.

Answer

Add a global secondary index that duplicates the primary key so reads can be spread across two indexes.

Answer

Increase the provisioned RCUs and enable auto scaling to keep up with the higher read volume.

Question 14

A developer's Lambda function reads customer profile items from a DynamoDB table using Query. Each item is roughly 40 KB and contains many nested attributes, but the function only needs the customer's name and email for a notification workflow. CloudWatch shows the function is consuming far more read capacity than expected and DynamoDB read costs are high. Which change most effectively reduces the consumed read capacity for these queries?

Accepted Answer

Add a ProjectionExpression to the Query request that returns only the name and email attributes.

Answer

Switch the Query to a Scan operation with a FilterExpression on name and email.

Answer

Enable a strongly consistent read on the Query to reduce the number of retries.

Answer

Increase the Limit parameter on the Query to fetch more items per request.

Question 15

A gaming application computes a global leaderboard by running an expensive aggregation query against an Amazon RDS database. The query runs on every page load, and during peak hours the database CPU is saturated, causing slow response times for all users. The leaderboard only needs to be accurate to within about 60 seconds. A developer wants to reduce database load with minimal code changes. What is the most effective approach?

Accepted Answer

Cache the computed leaderboard result in Amazon ElastiCache with a 60-second TTL, and serve subsequent reads from the cache until it expires.

Answer

Add a read replica to the RDS instance and direct all leaderboard queries to the replica.

Answer

Enable DynamoDB Accelerator (DAX) in front of the RDS database to cache the query results.

Answer

Increase the RDS instance size to a larger instance class to handle the peak query volume.

Question 16

A developer runs a product catalog service backed by Amazon RDS and uses Amazon ElastiCache for Redis with a lazy-loading (cache-aside) strategy. Users frequently report seeing outdated product prices for several minutes after an admin updates them in the database. The team needs to ensure the cache reflects price changes immediately after a database write, while keeping the cache populated. Which change best addresses this?

Accepted Answer

Add a write-through step so the application updates (or invalidates) the cache entry as part of every database write.

Answer

Increase the TTL on all cached items so entries stay in the cache longer and reduce database load.

Answer

Switch the ElastiCache cluster from Redis to Memcached to enable automatic cache invalidation on write.

Answer

Enable a larger node type for the ElastiCache cluster to increase read throughput and cache freshness.

Question 17

A developer's Lambda function connects to an Amazon RDS PostgreSQL database. Under moderate load, X-Ray traces show that each invocation spends 200-400ms establishing a new database connection, and the RDS instance is reporting an unusually high number of active connections that occasionally exhausts the connection limit. The developer wants to reduce both the per-invocation latency and the connection count without changing the database engine. Which change is the MOST effective?

Accepted Answer

Move the database connection creation outside the handler function so the connection is initialized once per execution environment and reused across invocations.

Answer

Increase the Lambda function's memory allocation so the connection is established faster on each invocation.

Answer

Set the Lambda function's reserved concurrency to a very high value so more environments are available to handle connections.

Answer

Open and close the database connection inside the handler on every invocation to ensure connections are always released.

Question 18

A developer maintains a Node.js Lambda function that makes many outbound HTTPS calls to a downstream REST API during each invocation. X-Ray traces show that a significant portion of each request's latency is spent establishing new TLS connections for every call. The function reuses the same HTTP client library across calls within an invocation. What is the MOST effective change to reduce this connection-establishment overhead?

Accepted Answer

Enable HTTP keep-alive on the HTTPS agent so TCP/TLS connections are reused across requests instead of being renegotiated each time.

Answer

Increase the Lambda function's memory allocation to speed up the TLS handshake computation.

Answer

Move the function into a VPC with a NAT gateway so outbound connections are pooled by the gateway.

Answer

Enable provisioned concurrency so the TLS sessions remain warm between invocations.

Question 19

A developer notices that a Lambda function backing a low-traffic REST API occasionally returns responses in over 4 seconds, while most requests complete in under 200 ms. AWS X-Ray tracing is enabled. When inspecting the slow traces in the X-Ray console, which segment detail most directly confirms that the latency is caused by a cold start rather than slow downstream calls?

Accepted Answer

An Initialization subsegment appearing in the function segment on the slow traces but not on the fast ones

Answer

A high Duration value on the DynamoDB downstream subsegment

Answer

A throttling error annotation on the API Gateway segment

Answer

A large gap between the API Gateway segment start time and the Lambda dwell time in the service map

Question 20

A data-processing Lambda function is configured with 128 MB of memory. It is CPU-bound and currently runs for about 8 seconds per invocation. The team wants to reduce both execution time and overall cost. After using AWS Lambda Power Tuning to test several memory settings, they find that at 1024 MB the function completes in about 1 second, and the total invocation cost is lower than at 128 MB. What is the primary reason increasing memory reduced both latency and cost in this case?

Accepted Answer

Lambda allocates CPU proportionally to memory, so more memory gave the CPU-bound function more compute, cutting duration enough to offset the higher per-millisecond price.

Answer

Higher memory tiers automatically enable provisioned concurrency, which eliminates cold starts and reduces billed duration.

Answer

Increasing memory reduces the number of billed invocations because Lambda batches more work into each request.

Answer

Lambda charges a flat rate per invocation regardless of memory, so any duration reduction is pure cost savings.

More DVA-C02 practice