Exam Overview
| Exam Code | Cloud-Deployment-and-Operations |
| Exam Name | WGU Cloud Deployment and Operations |
| Vendor | WGU |
| Total Questions | 67 |
| Last Updated | 26-Oct, 2026 |
Why CertsPoint?
- Latest & Verified Questions
- Detailed Explanations
- Regular Updates
- Full-Length Practice Tests
- Boost Your Exam Confidence
PDF Study Materials
Downloadable guides and resources
Practice Tests
Full-length tests with real exam experience
Progress Tracking
Monitor your performance and improvement
Study Anywhere
Access on desktop, tablet, and mobile
Secure Payment
Privacy Protected
Money Back Guarantee
Free Cloud-Deployment-and-Operations Practice Questions
Explore 5 free practice questions for the WGU Cloud-Deployment-and-Operations exam and review the answers and explanations to support your preparation.
-
Question 1
Multiple answers(A cloud engineer needs to notify the response team whenever a high-security web server responds
with a 403 Forbidden error. Which two steps can enable this functionality? Choose 2 answers.)-
Option A:Correct
Define a metric filter for Apache logs in CloudWatch
-
Option B:Correct
Bind a Lambda function to an Apache process
-
Option C:Correct
Create an alarm for the metric filter to deliver alerts using Amazon SNS
-
Option D:Correct
Send alarms from the Lambda function using Amazon SQS
Show Answer & Explanation Hide Answer & Explanation
Correct Answer: A, C
Explanation
To notify a response team when a high-security web server returns a 403 Forbidden error, two key
steps are required. First, define a metric filter for Apache logs in CloudWatch to detect the 403 error
code within the log data. This involves setting up a filter pattern to match "403" in the Apache access
logs. Second, create an alarm for the metric filter and configure it to deliver alerts using Amazon SNS,
which supports email or other notifications to the response team. The WGU Cloud Deployment and
Operations Study Guide (Section 4.2, CloudWatch Logs and Alarms) confirms that metric filters and
SNS-integrated alarms are the standard approach for monitoring and alerting on log-based events.
Options B and D are incorrect as they involve unnecessary or unsupported configurations (e.g.,
binding Lambda to Apache or using SQS for alarms). -
-
Question 2
Single answer(Which action must be used to create a metric filter in the Amazon CloudWatch console?)
-
Option A:Correct
Enable an alarm
-
Option B:Correct
Define a trace
-
Option C:Correct
Select a log group
-
Option D:Correct
Specify a stream
Show Answer & Explanation Hide Answer & Explanation
Correct Answer: C
Explanation
To create a metric filter in the Amazon CloudWatch console, the first step is to select a log group from
which the log data will be analyzed. A log group contains log streams, and metric filters are applied
to the log data within these groups to extract metrics based on patterns. The WGU Cloud
Deployment and Operations Study Guide (Section 4.2, CloudWatch Logs) specifies that the process
begins by navigating to the CloudWatch console, selecting a log group, and then defining the filter
pattern. Actions like enabling an alarm, defining a trace, or specifying a stream are subsequent or
unrelated steps. -
-
Question 3
Single answer(How are custom metrics grouped in CloudWatch?)
-
Option A:Correct
Namespace
-
Option B:Correct
Service
-
Option C:Correct
Date
-
Option D:Correct
Value
Show Answer & Explanation Hide Answer & Explanation
Correct Answer: A
Explanation
In Amazon CloudWatch, custom metrics are organized and grouped using namespaces. A namespace
is a container for CloudWatch metrics that allows you to isolate and categorize metrics from different
applications or services. According to the WGU Cloud Deployment and Operations Study Guide
(Section 4.1, CloudWatch Metrics), each custom metric must be assigned to a namespace, which acts
as a unique identifier to prevent naming collisions and facilitate metric management. Options like
Service, Date, and Value are not used for grouping metrics in this context -
-
Question 4
Single answer(Which AWS solution can send email based on CloudWatch alarms?)
-
Option A:Correct
Simple Queue Service (SQS)
-
Option B:Correct
Simple Notification Service (SNS)
-
Option C:Correct
Amplify
-
Option D:Correct
Kinesis
Show Answer & Explanation Hide Answer & Explanation
Correct Answer: D
Explanation
Amazon CloudWatch alarms can trigger notifications when a metric breaches a defined threshold.
The AWS solution designed to send emails based on these alarms is Amazon Simple Notification
Service (SNS). SNS supports sending notifications via email, SMS, and other protocols when
subscribed endpoints are triggered by CloudWatch alarms. The WGU Cloud Deployment and
Operations Study Guide (Section 4.3, Monitoring and Alarms) states that SNS is the primary service
for delivering notifications from CloudWatch, allowing users to configure email subscriptions for
alarm states. Other options like SQS, Amplify, and Kinesis are not designed for this purpose. -
-
Question 5
Single answer(Which CloudWatch metric filter includes log events with the word ERROR but excludes log events
with the word WARNING?)-
Option A:Correct
?ERROR ?WARN
-
Option B:Correct
"ERROR" WARN
-
Option C:Correct
ERROR -WARN
-
Option D:Correct
ERROR WARN
Show Answer & Explanation Hide Answer & Explanation
Correct Answer: C
Explanation
A metric filter in Amazon CloudWatch Logs can be used to search for specific terms in log data and
create metrics based on the matches. To include log events with the word "ERROR" and exclude
those with the word "WARNING," the correct syntax involves using a filter pattern with a positive
match for "ERROR" and a negation for "WARNING." The correct pattern is "ERROR -WARN," where
the minus sign (-) indicates exclusion of log events containing "WARN." According to the WGU Cloud
Deployment and Operations Study Guide (Section 4.2, CloudWatch Logs), metric filters use a patternbased
syntax where terms are included or excluded using positive matches and the negation
operator (-). This ensures that only logs with "ERROR" and without "WARN" are processed into the
metric. -