3.2.2. Choosing Storage Products and Storage Classes
💡 First Principle: Cloud Storage isn't one product with a fixed cost — the same bucket's storage class is a dial between retrieval speed/cost and storage cost, and choosing correctly means matching that dial to how often the data will actually be accessed, not how important the data is.
Cloud Storage holds unstructured objects (files, backups, media) with four storage classes: Standard (frequently accessed, no minimum storage duration, highest storage cost), Nearline (accessed roughly monthly or less, 30-day minimum), Coldline (accessed roughly quarterly or less, 90-day minimum), and Archive (accessed roughly once a year or less, 365-day minimum, lowest storage cost but highest retrieval cost). Filestore provides managed NFS file storage for workloads needing a traditional shared filesystem. Google Cloud NetApp Volumes and Google Cloud Managed Lustre serve more specialized enterprise and high-performance-computing filesystem needs, respectively, where Filestore's feature set isn't sufficient.
⚠️ Exam Trap: Nearline, Coldline, and Archive are separate storage services from Cloud Storage — a natural-sounding but incorrect assumption. They are storage classes assignable per object or as a bucket default within the same Cloud Storage service, not distinct products with different APIs — retrieving an Archive-class object uses the exact same API call as a Standard-class one, just with a different cost and latency profile.
Reflection Question: A company needs to retain compliance backup data for seven years, accessed maybe once if ever during an audit. Which storage class minimizes cost for this pattern, and what's the tradeoff if that data unexpectedly needs to be retrieved early?
