Topic 1 of 4
Many applications don't need the rigid structure of a relational database: they store images, videos, backups, or files to analyze. Azure Blob Storage is the service for that, and data lakes are built on it. The exam asks about blob types, access tiers, and what Data Lake Storage Gen2 adds.
Contents
A service for storing massive amounts of unstructured data in the cloud as binary large objects, or blobs.
Blobs are an efficient way to store files in a format optimized for the cloud, and applications read and write them with the Azure Blob Storage API. Azure Storage is a core Azure service that also offers file shares (Azure Files) and key-value tables (Azure Table Storage).
Storage account: The Azure resource that holds the data.
Container: Groups related blobs. Who can read and write is controlled at the container level.
Blob: The object itself: an image, a video, a file.
| Type | How it's organized | Maximum size | Best for |
|---|---|---|---|
| Block blob | A set of blocks of up to 4,000 MiB each | 190.7 TiB (4,000 MiB × 50,000 blocks) | Large, discrete binary objects that change infrequently |
| Page blob | Fixed 512-byte pages, random reads and writes | 8 TiB | VM virtual disks |
| Append blob | Blocks of up to 4 MiB added only at the end | Just over 195 GiB | Append operations, such as logs |
In a block blob, the block is the smallest unit that can be read or written. In an append blob, existing blocks can't be updated or deleted.
They balance access latency and storage cost:
Hot
The default tier, for frequently accessed blobs. Stored on high-performance media.
Cool
Lower storage cost and higher access cost than Hot. For infrequently accessed data. Minimum 30 days to avoid early deletion penalties.
Cold
For data rarely read but needing fast retrieval: short-term backups, disaster recovery. Minimum 90 days.
Archive
The lowest storage cost and the highest latency. For history that mustn't be lost but is rarely used. Minimum 180 days.
Rehydration
In Hot, Cool, and Cold, reads take milliseconds. Archive blobs are effectively offline: to read them you change them to Hot, Cool, or Cold and wait for rehydration to complete, which can take up to 15 hours.
It's common to create a blob in Hot and move it to Cool once it's accessed less; it can also move back from Cool to Hot.
A lifecycle policy can automatically move a blob from Hot to Cool, then to Cold, and finally to Archive as it ages and is used less. The rule is based on the number of days since last modification. It can also delete outdated blobs.
Hot
→ Cool
→ Cold
→ Archive
→ Delete
| Option | Where copies are kept | Protects against |
|---|---|---|
| LRS | Three copies in a single datacenter | Hardware failures |
| ZRS | Copies across three availability zones in the primary region | A zone going down |
| GRS / GZRS | Asynchronous replication to a secondary region hundreds of miles away | Regional disasters |
| RA-GRS / RA-GZRS | Like GRS/GZRS, with read access to the secondary | Reading before a failover |
A cloud-scale data lake solution built into Azure Storage. It combines Blob Storage's scalability and cost control (tiers and lifecycle included) with a hierarchical file system compatible with major analytics systems.
One-way only
Upgrading an account to the hierarchical namespace can't be undone: afterward you can't revert to a flat namespace.
Every Microsoft Fabric tenant automatically provisions OneLake, a unified data lake for the whole organization that is built on Azure Data Lake Storage. It supports the ADLS Gen2 APIs, so existing applications and services can connect to it. Domain D4 covers it as part of Fabric.
| Situation | Answer | Why |
|---|---|---|
| Store large binary files that rarely change, such as videos | Block blob | Large, discrete objects. |
| Virtual machine disks | Page blob | Random reads and writes on 512-byte pages. |
| Logs that only get appended to | Append blob | Blocks are only added at the end. |
| Data queried daily | Hot tier | High-performance media, the default tier. |
| Backups read rarely but needing fast retrieval | Cold tier | Cheap, with millisecond reads. |
| History almost never read that can take hours to retrieve | Archive tier | The cheapest, offline until rehydrated. |
| Automatically move blobs to cheaper tiers as they age | Lifecycle management policy | Based on days since modification. |
| A data lake with real folders and per-file permissions | Data Lake Storage Gen2 | Hierarchical namespace and POSIX ACLs. |
| Keep reading data even if a whole region fails | RA-GRS or RA-GZRS | Read access in the secondary region. |
❓ Which blob type does Azure use for virtual machine disks?
Page blobs, optimized for random reads and writes.
❓ Can you read an Archive blob instantly?
No. You must change it to Hot, Cool, or Cold and wait for rehydration, which can take up to 15 hours.
❓ What is the default access tier?
Hot.
❓ Do Blob Storage folders let you apply per-folder permissions?
Not by default: they're virtual. With Data Lake Storage Gen2's hierarchical namespace you can, using POSIX ACLs.
❓ Can you turn off the hierarchical namespace after enabling it?
No. It's a one-way process.
❓ What does a lifecycle policy use to move blobs?
The number of days since last modification.
❓ What are the minimum days for Cool, Cold, and Archive?
30, 90, and 180 days respectively, to avoid early deletion penalties.
Did you get this topic?
Test what you just learned
A company must keep audit logs it almost never queries for 7 years. If it needs them, it can wait several hours. It wants the lowest storage cost. Which tier does it use?
Original questions, each with a full explanation.
Start a mock exam