Data tiering

Data tiering is placing each file on storage matched to how often it is read and how much it is worth to the business. Frequently read data sits on the fastest tier, rarely read data on slower and less costly storage, and data kept for the record on archive storage. The decision is economic, and it depends on knowing access patterns.

Key takeaways

Tiering is an economic decision: fast storage for what is used, low-cost storage for what is kept.
Business attributes such as project and owner can drive tiering, not only age.
File-level tiering moves whole files, visibly and reversibly.

Why does data tiering matter?

Because most data is read rarely after its first months, yet it often stays on the fastest and most expensive storage. Tiering moves it to storage that matches how it is used, so fast storage is kept for active work. The result is lower cost per terabyte and fewer emergency purchases.

Does data tiering always mean moving files to the cloud?

No. Tiers can sit anywhere, and many organizations never use the cloud for tiering at all. Common lower tiers include:

Lower-cost disk in the same data center, such as a capacity NAS behind a fast flash system.
On-premises object storage, which suits large data that is read rarely.
Tape, for long-term archives that are almost never read.
Cloud storage, including archive tiers that cost little to keep but more to read back.

The right tier depends on how often the data will be read, how fast it must come back, and any rules on where it may be stored.

Can data be moved back to a fast tier if it is needed again?

Yes. Moving data back up, often called a recall, is a normal part of tiering. It can happen on request, when a user or application opens a file that was moved, or by rule, such as when a project is reopened. How long it takes depends on where the data went: files on lower-cost disk come back in seconds or minutes, while tape and cloud archive tiers can take hours and may charge retrieval fees. Planning for recalls up front, by keeping data that might be reopened on a faster lower tier, avoids most surprises.

Can tiering use business attributes instead of file age?

Yes, when the catalog holds them. A tiering rule can use fields such as project, status, or retention class, so a completed project that has not changed in a year moves to the archive tier even if someone opened one of its files last week. Age and last access still matter, and business attributes make the decision closer to what the data is worth. The storage system moves the files, and the catalog decides which files move.

What is the difference between file-level and block-level tiering?

Block-level tiering moves blocks within a storage system and is invisible to users and to any catalog. File-level tiering moves whole files between systems and tiers, so each move is visible, attributable, and reversible. Only file-level tiering can be driven by attributes such as project, owner, or retention class.

What is archival storage, and when should data be archived?

Archival storage is low-cost storage for data that must be kept and is rarely read. The right moment to archive is when a file has passed its active life and still has a retention obligation or a plausible future use. Deciding that from evidence rather than instinct is the entire difficulty.

What is hierarchical storage management (HSM)?

Hierarchical storage management is an older approach to tiering in which a system moves files from fast disk to slower, less costly media, such as tape, and leaves a small placeholder, or stub, in the original location. When someone opens the stub, the system brings the file back. HSM saves capacity, but stubs can confuse backups, migrations, and search tools, which is why many organizations now prefer file-level tiering that moves whole files visibly.

What is bit rot?

Bit rot is the slow corruption of stored data as storage media degrade, so a file that has not been read for years may no longer read correctly. It is also called data decay. Storage systems guard against it with checksums and scrubbing, which reread data and repair it from redundant copies. A catalog of last access dates shows which files have gone longest without being read, which is a useful starting point for deciding what to verify.

What is the difference between data tiering and cloud tiering?

Data tiering is the general practice of placing data on storage that matches how it is used, wherever that storage sits. Cloud tiering is one kind of data tiering, where the lower tier is cloud storage. The same decisions apply to both, such as which files to move and when. Cloud tiering adds a few of its own, such as egress and retrieval fees, network time, and where the cloud region is located.

What is the difference between data tiering and data mobility?

Data tiering moves data between storage tiers based on how it is used, usually down to lower-cost storage as it cools, and sometimes back up when it is needed again. Data mobility is broader: moving data in any direction, for any reason, such as sending a project to a cloud service for processing, sharing it with another site, or changing storage vendors. Tiering is one use of data mobility, and both rely on data movement, the act of copying files from one place to another.