I achieved a provisional pass on the GCP Developer Professional Exam. This is my observations out of the exam. This isn't a tied to specific questions, if something was just one question, it wouldn't really stick out. This is about trends and themes across the exam.
- Sathish Vj's notes are quite good, I recommend reviewing all the docs he links to.
- Understand LB deployments- GCP has a lot of entities involved in an LB deployment (instance templates, managed instance groups, backends, etc). Be familiar with how to provision the full stack, as it were.
- Remember the IP addresses and firewall rules needed for health checks. I didn't have to recognize the IPs from a range of conflicting values, but know what those IPs are, and why they matter.
kubectl: know the subcommands. Understand how logs work.- IAM Permissions: Understand how service accounts work across projects. I had a few questions about permissions for resources in project and a b interacting.
- Datastore: Be familiar with best practices- how to use transactions, when to use ancestor queries.
- Bigquery:
- batching- I wasn't explicitly asked about batching, but one of the solutions was based on if batching made sense, so don't just look for the word.
- DML: know what combinations can run concurrently.
- GDPR: Know, at a high level, what design is needed to implement compliant apps.
- GitOps: Understand how to implement a pipeline from source control to a build tool to a container registry. Most of the time, this revolves around GCP products, but occasionally Docker Hub and Jenkins were mentioned, so know what they are analogous to.
gcloudvsgsutil:gsutilis the storage utility. If its not interacting with a bucket, you won't need this. Also pretty sure they made up some commands line names.- Troubleshooting: What should you do when a vm is unresponsive to ssh? Its covered in the study guide- check the serial port output, and attach the disk to another vm to diagnose further.
- IAP came up a few times- I didn't need to know all the support mechanism, but all of them product a JWT.
- Dataproc: I got a few questions, so know what it is, what the use cases are, and how to manage/import/export data.
- Case study: I got about 10 questions on the case study. Know the data bases- what to use when (spanner vs sql vs data store vs big query)
- This exam was more focused on remembering specific details than the PCA- which often was 'Here are four right choices, pick the most correct one. Definitely be able to rattle off use cases and limitations for all the services (I know, good luck).
I hope this helps. If anything in the guide is inaccurate, please do open an issue or a PR!