I'm literally just going to rapid-fire list some of the things that are tested on GCP exams that people get wrong.
The three phases of a Migrate to Virtual Machines migration are: Discovery (finding and cataloging source VMs), Replication (continuously replicating VM data to GCP while the source VM is still running), and Cutover (performing the automated switch to move production workloads to Compute Engine).
PCI DSS stands for Payment Card Industry Data Security Standards.
Cloud KMS can be used to provide the secure, centralized management of encryption keys that are used to protect cardholder data at rest and in transit.
TCP 443 (HTTPS) is used for communication with the Dataflow service control plane, API requests, and metadata exchange.
It's for the nodes' external communication and is non-negotiable.
For streaming jobs, or batch jobs that require shuffling data between workers, TCP 12345 and 12346 are also necessary.
True! They are not universal, but they automatically kick in once criteria are met.
Inspect logs at the pipeline and the worker level to identify specific stages in the pipeline with higher latency. Look for resource limitations, data skew, or dependencies in the pipeline that might be causing delays at particular stages. Professional Data Engineer.
Balanced = general workloads, balance of performance and cost
Standard = most basic disk, cheapest, low I/O workloads and backups
SSD (solid state drive) = high performance and IOPS for databases and analytics
Extreme = maximum performance, highest cost, for demanding databases