AZ-305 Practice Question 44: Analytics Storage Foundation
Microsoft Azure Solutions Architect Expert (AZ-305) · Test 5 · question 4 of 10
A manufacturer is building a data lake that Azure Databricks and Synapse Analytics will query. The design must satisfy the following:
- store several petabytes of raw and curated files in open formats, at object-storage prices, with lifecycle tiering of cold data, - let Spark connect through the drivers it already uses for Hadoop-compatible file systems, - grant different engineering teams permissions on individual directories rather than on the whole container, - rename and delete curated directories as a single atomic operation, because the nightly job reorganizes large folder trees.
What should the design use?
Show answer and explanation
Answer:
Data Lake Storage is a set of capabilities on top of Blob Storage that you unlock by enabling the hierarchical namespace, and each requirement maps to one of them. Hadoop distributions include the Azure Blob File System (ABFS) driver, so Spark and Presto read the data directly. The hierarchical namespace organizes files into real directories, which makes renaming or deleting a directory a single atomic metadata operation instead of an enumeration of every object sharing a prefix. Its access control model supports POSIX ACLs set at the directory or file level alongside Azure RBAC. And because it is built on Blob Storage, it keeps blob pricing, access tiers, and lifecycle management.
Option A gives the price and the scale but neither of the other two requirements. Without a hierarchical namespace a "directory" is only a name prefix, so a rename is a copy of every object underneath it, and permissions cannot be set per directory.
Option B is the wrong storage class for analytics at this scale. Azure Files is an SMB/NFS file share priced and sized for shared application storage, it tops out well below several petabytes, and it has no HDFS-compatible driver for the analytics engines.
Option C is a relational database, not a lake. It cannot store raw files in open formats, its storage costs far more per terabyte than object storage, and it offers no directory-level ACLs or Spark file-system access.
Reference: Azure Data Lake Storage overview