The AWS Solutions Architect exam has 65 questions and you will have 130 minutes to complete it, this is just over two hours, if you do the math, you'll have about two minutes to answer each question. Some questions read faster than others. Likewise, there will be questions that are simply put easy to answer. The two minute per question is just an average. However, if you're prepared, none of the questions will take that long to answer.

Each of the 65 questions falls into one of four domains. A passing score is 720 points out of a possible 1,000. This is a scaled score, that is internally AWS gives some questions more weight than others. Also inside the exam guide is a paragraph that has the heading Unscored Content. It says, "Your examination may include non-scored questions that are placed on the test to gather statistical information."

Non-scored means exactly that, right or wrong they have no impact on whether or not you'll pass the exam. You will not know which ones they are, why they're included or if you got them correct or not. If you've spent any amount of time with AWS you have either attended or watched videos of re:Invent. New services are often referred to as an MVP, a Minimally Viable Product, then over time these services get more features and become more robust.

A passing score of 720 for the certified developer exam is similar though instead of being minimally viable, it's more like you're minimally qualified. Minimally qualified means that you have enough skills to get the job done and be recognized as such. I think of it like television shows and movies that feature doctors they sometimes talk about their oath to first, do no harm. That's at the core of what being minimally qualified means. You can do the work without doing harm to yourself, others or the data in your care.

Let's look at that idea of doing the work. The exam is designed to test real-world AWS knowledge and concepts that are relevant and current. It will avoid trivial knowledge like button placement or things that would normally require consulting documentation when needed. Instead it tests common and important tasks that are relevant to a developer's role. This includes evaluating the ability to solve problems improve performance, troubleshoot, or improve security.

In essence, the exam tests competency and not theoretical or trivial knowledge. By trivial knowledge, I mean technical jargon, local terminology, workplace specific verbiage, and complex language. These things do not measure ability and are not found on the exam. If the exam tests competency, how does it do that? The answer is actually found in the exam guide. It is important to review and understand the exam guide because each question is written to measure your knowledge of a specific skill against one of its stated objectives.

What most people think of as a test question really has three parts, a scenario a specific question about that scenario and a set of possible answers. The scenario is officially known as a STEM and is the setup for a question, its purpose is to introduce the concept being evaluated. STEMS include all of the details needed to make an informed decision about the correct answer.

The question that follows the STEM focuses on a single concept or aspect of AWS that is directly related to the exam guide. You are only tested on one thing at a time. That is a question could ask for a solution that is highly available or cost-effective, but not both. Common question themes include being highly available, cost effective, the most secure, performance improvement and the least amount of downtime. You will not be asked for the best or easiest solution because what's best for you might be the worst for someone else. Instead, questions often have qualifiers that identify a situation, examples include which of these options is the most cost-effective? Which action has no downtime? Which has the least amount of latency?

AWS exam questions focus on the use of its services and features, that is you're being tested on whether or not you have the knowledge and skills needed to implement solutions. This means you will not need to memorize the tails like performance metrics for EC2 instances, service limits, the syntax for the AWS CLI and prices for various services. Thinking back to what I said earlier this is trivial information and are things that could easily be looked up at the time of implementation, they are not worthy of being on the exam.

Before I move on to a description of the responses, there's one other thing I want to share regarding the questions, they are either open-ended or closed. An open-ended question does not end with a question mark. These are typically fact-based statements that are cut off mid sentence and have response options which complete the sentence in a grammatically correct way. For example, AWS is an acronym that stands for Amazon Web A, systems, B, services C, solutions or D, structures? If you don't know this, you might be in the wrong place.

In contrast to this, a closed question is fully contained. That is it presents a complete scenario and asks a specific question related to it. The easiest way for me to think about it is that a closed question ends with a question mark. Also, you will not have responses that are combinations of others responses. You know, A, C and D, but not B A and D, but not C or B, B and C only. That's probably everything important about the questions. Now let's get to the responses.

The list of possible responses include a correct answer and a number of incorrect answers. In the world of exam creation the correct answer is also called the key and incorrect answers are referred to as distractors. Distractors are incorrect within the context of the question, as you take AWS exams, you'll find that while these options are incorrect, they are still plausible that is, there is some truth to them. They could even be true within another context.

Sometimes what makes a distractor plausible is a word or phrase that sounds as if it's correct however, if you know the subject being tested it can be wrong to the point of being amusing. It's like a honeypot, it will be attractive to those that are not fully knowledgeable. This is the type of question that will trap someone that fails to have the skills of a minimally qualified person.

There are two types of questions on the exam multiple choice and multiple response. Multiple choice questions have one correct response and three incorrect responses. Multiple response questions have two correct responses out of five possible options. These are the only types of questions on the exam. There are no true/false, matching, drag and drop short answer or fill in the blank questions.

Now, this might seem strange to say out loud but to get a question correct you have to select the response or responses that best completes the statement or answers the question. The key word in that sentence is best. I've seen more than one question in AWS exams that had correct answers that were not close to how I would do things in the real world. If there's a trick, it's to recognize what's possible and eliminate those answers that are impossible or absurd. For example, a question about updating a file stored in S3 requires you to understand that objects in S3 are immutable, they cannot be changed, only replaced. So a correct answer has to include the concept of creating a new object, anything else is wrong and can be eliminated.

I've been asked what happens if you skip a question and don't answer it, unanswered questions and incorrect answers are scored the same, they're incorrect there is no penalty for guessing. Regarding multiple response questions, they're either 100% correct or they're wrong there's no partial credit. You can mark questions for review, when you finish the exam you'll be prompted to review any questions you've marked.

The people responsible for creating the AWS certification exams work hard to ensure there are no teaching statements in the exam. That is there's no language in exam questions that describes any AWS service or feature. If there was, it would be possible for one question to actually answer another one. For example, this is a teaching statement. Your organization is using AWS Lambda, a compute service that will let you run code without provisioning or managing servers. If another question on the same exam asked which service would you run code without provisioning or managing servers? People that don't know the answer could still answer it correctly because the exam itself provided the answer.

I have never seen any teaching statements on the AWS exams I've taken, for this reason even though it's possible to flag questions for later review, I don't it's rare that any given question has helped me answer another one even by triggering a memory. When I'm done, that's it, I end the exam. When you're done, you'll be asked if you're sure. It will prompt you, I think, three times before the exam is ended. You'll be prompted, are you sure? Yes, I am, click. Then the exam will prompt you again, are you really sure? There's no turning back. Yes, I'm sure, why are you torturing me? Click, it will then prompt, are you really, really sure? This is a nice exam you have here, it would be a shame if something bad happened to it. Yes, I'm sure, please stop taunting me I need this to be over, click.

Okay, maybe those aren't the actual prompts it's just how I felt after nearly every exam experience. After that third prompt, you might think that this is when you discover whether or not you've passed the exam, you'd be wrong. Then you'll be given a survey. I've taken these exams for years and this part is the most annoying. I wanna know if I've passed however, the people at AWS want to know if I felt the exam was an accurate representation of my skills. The question I think, is impossible to know without knowing whether or not I passed. My feedback has been the same for years. However, the exam experience has not changed.

The Mechanics of the Exam