I just passed the second of my GCP certifications. After about 2 years of designing data architectures for solutions on GCP cloud at a large financial organisation as a consultant I thought it was necessary to formalise my experience. The certifications also look good on my profile for would be clients and employers.
This short blog is about the training materials I used and some tips for others interested in pursuing the same certifications. It is also a reference for me because Google certifications expire after 2 years!
1. Linux Academy – https://app.linuxacademy.com/
- Google Cloud Certified Professional Cloud Architect
- Google Could Certified Professional Data Engineer
Notes: This is my first choice for those studying of these certifications. The course content is presented without the Google marketing I noticed on Coursera. It costs just under $50 for a monthly subscription that covers all courses available on site, with access to the mobile version where one can download course content to watch offline on the go. Subscriptions also include unlimited access to Google Labs. My favourite feature on the site is the reference to foundational courses to prepare students for work and the examination. The tests at the end of the courses simulate the real exam and passing them indicate the student is nearly prepared to write the exam.
2. Coursera – https://www.coursera.org
- Architecting with Google Compute Engine
- Data Engineering, Big Data, and Machine Learning on GCP
Coursera is the Google recommended training provider for preparing for their certification examinations. But not for me. The instructors were all Google employees who were out to market the company’s services first before delivering the courses. I don’t think the content is sufficient to pass the exams. The course delivery sounded “mechanical” in places, like they were read from slides. On the plus side, Coursera’s courses include Qwiklabs hands-on-labs and tests after each module. Coursera offers monthly subscription although students can audit courses for free i.e. take course without access to tests of labs, a good way get started. Another plus.
3. A Cloud Guru – https://learn.acloud.guru/dashboard
- Google Certified Associate Cloud Engineer
- Google Certified Professional Cloud Architect 2020
A Cloud Guru are stronger on the AWS courses than Google although I found the Associate Cloud Engineer course useful after taking courses on Coursera. The Professional Cloud Architect course on A Cloud Guru was total waste of my time! For those without GCP experience I suggest taking the Associate Cloud Engineer course on A Cloud Guru and if you can, write the exam before attempting the PCA certification.
Key topics / areas of focus
From what I can recall, the questions I faced in the exam were from the following areas:
The advice I got during preparation was to expect circa 30% of the questions from the cases studies:
- MountKirk Games
In the exam, I thought the number was closer to 25% of the questions which comes to about 13 questions in all. My advice: read and understand the case studies thoroughly, sketch the as-is-state and the target state by yourself for each scenario, applying the Google recommended tools, services and approaches before writing the exam. Understanding the business and technical requirements for each case studio helped me in small measure.
General GCP Services
Very few question were theoretical, most were scenario-based. On the whole, the expectation is to have all round understanding of GCP services. The only thing I didn’t see was Site Reliability Engineering (SRE), maybe it’s not part of the exams…
Here is what I recall:
Identity & Access Management: Organisations, Folders, Projects, and resources inside projects. Understand inheritance of the resource hierarchies with respect to IAM, Identity Federation (GCDS) and specifically roles and access control for BigQuery.
- Understand when to use table partitioning and when to configure retention and to apply expiration on tables and partitions
- Understand use of BigQuery for sharing analytics with third parties
Cloud Spanner: This is a bonus; question always has keywords such as “global consistency”, “transactional database”, etc.
NoSQL Data Stores
- Understand the differences between BigTable and Cloud DataStore and when to use either.
- Understand indexing on Datastore and the nuances of data stored in it.
Understand to the classes and when to use them: Nearline vs Regional vs Multi-regional vs Coldline
- Understand lifecycle management specifically how to apply the lifecycle JSON to a GCS bucket.
- Also understand the use of GCS to share data with third parties, especially those without GCP accounts i.e. signed URL with expiration.
Cloud Pub/Sub: I only recall this came out under the case study question but one should know when to use it. Also understand the IAM policies required for Cloud Engine to write to or subcribe to Pub/Sub topics
Cloud DataFlow: Almost all questions I recalled that included DataFlow also had Pub/Sub as a part of the answer. So understand the “Kappa” architecture or stream processing i.e. Pub/Sub DataFlow BigQuery/BigTable.
Cloud Storage Transfer Service / Cloud Transfer Appliance: Understand the difference between these two services and when to use one over the other. Understand the what “rehydration” means and when it is used.
Networking (VPC, VPN, Subnets & Connectivity): Luckily, I had few questions on this; my weakest area! But there were. a couple of question on setting VPN between on-premises and Google. I remember one of was about the type of connectivity to support a 20 Gbps speed to Google.
Reg & Compliance: There was one or two question about GDPR and data retention to meet regulations. Understand that using Google services do not mean that a solution is compliant, the design approach is what makes it compliant.
- Google Kubernetes Engine: a few question, recal on where I selected using gcloud to set up the cluster and kubectl to set up the deployment
- Only one question about Cloud DataPrep, using it to clean data and storing the data in Cloud Storage
- StackDriver: there were a couple of questions on StackDriver. One was about using StackDriver for analysing performance of an application directly; another was about storing Stackdriver logs in Google Cloud Storage for long term archiving. Understand how to expose logs for real time analysis using Pub/Sub as a sink.
• Make sure you’re consistently scoring the 90% and above in the preparatory tests on any of the training sites e.g. on Linux Academy tests. Review the explanations for the questions you miss and don’t hesitate to challenge them if the answers provided are wrong. I did this a few times on Linux Academy and it was corrected. It is probably a sign that you’re prepared for the exam.
• Make sure you are consistently scoring 100% on the Google Cloud Practice exam.
• I found the Qwiklabs, Codelabs and Hands-on-Labs from training websites very useful to cement understanding of the concepts.