Design data storage solutions for semi-structured and unstructured data

Bellhaven Architects has 15 TB of project files on an on-premises Windows file server accessed through SMB shares with NTFS permissions tied to its Active Directory domain. The firm is moving the files to Azure, wants users in three branch offices to keep a fast local cache of frequently used files, and wants to retain the existing NTFS access control lists. Administrative effort must be minimized.

Which storage solution should you recommend?

1. AAzure Files SMB shares with Azure File Sync and on-premises Windows Server cache endpointsAzure Files supports SMB with identity-based authentication that preserves NTFS ACLs, and Azure File Sync turns the branch office Windows Servers into caching endpoints with cloud tiering, keeping the cloud share as the central copy.
2. BAzure Blob Storage with the NFS 3.0 protocol enabledNFS 3.0 on Blob Storage is for Linux workloads and does not support SMB, Windows ACLs, or Active Directory integration.
3. CAzure NetApp Files SMB volumesNetApp Files supports SMB and NTFS ACLs with high performance, but it provides no branch-office caching service equivalent to Azure File Sync and is a more expensive, higher-performance tier than the scenario requires.
4. DAzure Data Lake Storage Gen2 with hierarchical namespaceData Lake Storage Gen2 is optimized for analytics with POSIX-style ACLs over the Blob API; it is not an SMB file share for Windows users.

Tidewell Bank is designing an Azure Cosmos DB for NoSQL account that is replicated to three regions with a single write region. A customer who updates a profile and then immediately reloads it from the same session must always see their own write. Users in other regions may see slightly stale data. Tidewell wants the lowest-latency, lowest-cost consistency level that meets this requirement.

Which default consistency level should you recommend?

1. AStrongStrong consistency guarantees every reader sees the latest committed write, but it incurs the highest read latency and request unit cost across regions, which exceeds the requirement.
2. BBounded stalenessBounded staleness limits lag to a number of versions or a time interval; it also provides read-your-writes within the write region but costs more than session consistency and is typically chosen for global ordering guarantees.
3. CSessionSession consistency guarantees read-your-own-writes, monotonic reads, and consistent prefix within a client session at the cost of eventual consistency for other sessions, with lower latency and read cost than strong or bounded staleness.
4. DEventualEventual consistency is the cheapest but gives no ordering or read-your-writes guarantee, so a user could reload and see their old profile.

Ironvale Manufacturing stores quality inspection photos in an Azure Storage account in the West Europe region. The photos must remain available for read access if the entire West Europe region becomes unavailable, without waiting for Microsoft or Ironvale to initiate a failover, and must also survive the loss of a single availability zone within West Europe. Ironvale wants the lowest-cost redundancy option that meets these requirements.

Which redundancy option should you recommend?

1. AZone-redundant storage (ZRS)ZRS replicates synchronously across three availability zones in one region, which protects against a zone loss but not against a regional outage.
2. BGeo-redundant storage (GRS)GRS keeps a copy in the paired region but uses locally redundant storage in the primary region, so a zone loss in West Europe would make the data unavailable until a failover; the secondary is also not readable until failover.
3. CRead-access geo-zone-redundant storage (RA-GZRS)RA-GZRS combines zone-redundant replication in the primary region with asynchronous replication to the paired region and exposes a read-only secondary endpoint that applications can use immediately during a regional outage.
4. DGeo-zone-redundant storage (GZRS)GZRS provides both zone and regional durability, but the secondary region is not readable until a failover occurs, which fails the no-wait read requirement.

Sablewood Media stores video uploads in an Azure Blob Storage container. Videos are viewed heavily for the first 30 days, rarely in the following six months, and almost never afterward, but legal policy requires that they remain retrievable for five years and retrieval of an old video can take several hours. Sablewood wants to minimize storage cost with no ongoing manual work.

What should you recommend?

1. AStore all videos in the Hot tier and delete them after five years with a scheduled scriptKeeping rarely accessed data in the Hot tier pays the highest per-GB price for years, and a custom script adds maintenance.
2. BCreate a lifecycle management policy that moves blobs to Cool after 30 days, to Archive after 210 days, and deletes them after five yearsLifecycle management applies tier transitions and deletions automatically based on blob age, and the Archive tier offers the lowest storage price with retrieval times of up to several hours, which the policy accepts.
3. CCreate a lifecycle management policy that moves blobs to Cool after 30 days and keeps them there for five yearsCool is cheaper than Hot but considerably more expensive than Archive for data that is almost never read after seven months.
4. DUpload new videos directly to the Archive tier and rehydrate them on demandArchive blobs are offline; the heavy viewing in the first 30 days would require hours-long rehydration for every view and incur high retrieval charges.