Every company has data: photos, contracts, backups, logs. Azure offers several storage services for it, and each fits a different purpose. In this lesson you learn what they are, how you control costs and resilience through tiers and copies, and how data and entire servers get to Azure.

The running example is the made-up company Weidemann Druck GmbH. It has an archive of scanned orders, a file server in the office and a few old servers that are to move to the cloud.

The storage account and the storage services

The services of Azure Storage live in a shared container, the storage account. It is a shared storage pool in which you create the individual resources. Every account has an address of its own. Important: The redundancy setting applies to all services of the same account. If two kinds of data need different safety, you create two accounts.

The data services at a glance:

Service What it is Typical use
Blobs (Blob Storage) Object storage for large amounts of unstructured data such as text and binary files Images and documents for the browser, videos, backup and archive
Files (Azure Files) Managed file shares, reachable through the standard protocols SMB and NFS Replacing a file server, moving applications that use a file share to the cloud
Queues Message store for communication between parts of a program Decoupling tasks, such as "create thumbnail"
Tables NoSQL store for structured data without a fixed schema Flexible records such as address books or device information
Managed Disks Block storage, that is, virtual hard disks for virtual machines The drive of a server

Remember the rule of thumb: Files for all programs on the network are Files, bulk data for your own applications is Blobs, and the disk of a virtual machine is a Managed Disk. Azure Storage encrypts all data written to an account, and Azure takes care of maintenance and updates of the hardware.

Access tiers: What does it cost to keep data?

Not every file is needed every day. That is why there are access tiers for blobs. The cooler the tier, the cheaper the storing and the more expensive the access:

  • Hot: for data that is read or changed frequently. Highest storage costs, lowest access costs.
  • Cool: for rarely used data that should still be readable immediately. It should stay there for at least 30 days.
  • Cold: for data that is needed rarely but must be quickly retrievable. At least 90 days.
  • Archive: an offline tier for data that is hardly needed and for which a wait of hours is fine. At least 180 days.

Whoever deletes data before the minimum time has passed, or moves it to another tier, pays a prorated early deletion fee. An archived blob can't be read. It must first be brought back to an online tier (rehydration), and that can take up to 15 hours. You set the tier only for block blobs. A default for new blobs belongs to the account, and lifecycle management rules can move data into cooler tiers automatically.

Redundancy: How many copies, where?

Azure always stores several copies of your data. You choose how far they are spread:

Option What happens Survives
LRS (locally redundant) Copies in a single datacenter of the primary region Failure of a drive, server or rack
ZRS (zone-redundant) Synchronously in three or more availability zones of the primary region Failure of an entire datacenter
GRS (geo-redundant) LRS in the primary region, plus asynchronously to a far-away secondary region Failure of an entire region (after a failover)
GZRS (geo-zone-redundant) ZRS in the primary region, plus asynchronously to the secondary region Zone and region failure

With GRS and GZRS the secondary region can be read only after a failover. Whoever wants to read the data there beforehand takes the variants with read access, RA-GRS and RA-GZRS. Because the copy to the secondary region runs asynchronously, the most recent change can be missing after a total failure of the primary region. And remember: Redundancy protects against hardware failure, not against deletion. If a file is deleted, it disappears from all copies.

For archives there is one restriction: The Archive tier is supported only by accounts with LRS, GRS or RA-GRS, not ZRS, GZRS or RA-GZRS.

Storage account types

Microsoft recommends Standard general-purpose v2 as the default type. It supports Blobs, Files, Queues and Tables. Besides that there are premium types for special requirements: Premium Block Blobs for blobs, premium file shares on an SSD basis and Premium Page Blobs. The region and the access tier also affect the price. Microsoft lists the older types, such as general-purpose v1, as "legacy"; for new accounts, v2 is recommended.

Moving data

How does Weidemann Druck get its data to Azure? Depending on the case, there are different tools:

  • AzCopy is a command-line program that copies files between your computer and Azure Storage, also between two storage accounts. Good for scripts and one-time copies.
  • Azure Storage Explorer is a free application with a graphical interface for Windows, macOS and Linux with which you view and manage your storage data.
  • Azure File Sync turns a Windows file server into a fast cache of an Azure file share. The frequently used files stay on the server, the rest live in the cloud (cloud tiering). Ideal if the file server is to stay in the office.

For entire moves there are two more services:

  • Azure Migrate is the central hub for discovering, assessing and moving servers, databases and web apps to Azure. The service itself is free.
  • Azure Data Box is a device that Microsoft ships to you. You copy your data onto it, ship it back, and the data is loaded in the Azure datacenter. That pays off if the network is too slow, too expensive or not there at all.

The example of Weidemann Druck GmbH

The scan archive is large and hardly looked at anymore. It moves as blobs into the Cold or Archive tier. The file share in the office stays on site and is backed up and mirrored with Azure File Sync. IT discovers and assesses the old servers with Azure Migrate. Because the office has only a thin line, it loads the first large body of data with a Data Box.

Exercise 1

Match each case with a solution. Justify each in one sentence.

  1. A club wants to keep the photos of its annual celebrations for five years. They are almost never retrieved, and waiting a few hours doesn't matter.
  2. An application needs a file share that several virtual machines read and write at the same time.
  3. A web shop stores product images. If an entire datacenter fails, the images should stay reachable, and the data must not leave the region.
  4. A company has 80 terabytes of data and a very slow line.

Solution 1
  1. Blobs in the Archive tier. Rarely needed, waiting time not critical, lowest storage costs. Important: The account must use LRS, GRS or RA-GRS.
  2. Azure Files. A managed file share over SMB or NFS that several machines mount at the same time.
  3. Blobs with ZRS. Copies in several zones of the same region: The failure of a datacenter is survived, and the data stays in the region. The images are read frequently, so the Hot tier fits.
  4. Azure Data Box. The device arrives by freight carrier, the data is copied on site and loaded in the datacenter. The network isn't needed for that.

Further reading