In short
- Definition: The ability of a system or its data to persist over time without loss or degradation, despite failures.
- Goal: Make sure that once data is committed, it is not lost.
- Example S.M.A.R.T. requirement: 99.999999999% annual durability of stored objects, i.e., expected loss below 1 object in 100 billion per year.
1. Definition
As a non-functional requirement, durability describes the ability of a system, or of the data it stores, to persist over time without loss or degradation despite hardware failures, crashes or power cuts [1], [2], [3].
Durability protects against the machine. Backups protect against the human.Key takeaway
In databases, durability is the “D” of ACID: once a transaction is confirmed, it must survive a crash. In storage services, durability is expressed as the probability of keeping an object over a year. For example, Amazon S3 states that its standard storage is designed for 99.999999999% (eleven nines) annual durability [4].
Durability is not availability. Data can be perfectly durable but temporarily unreachable, and a highly available service can still lose data if writes are not safely persisted.
2. Goal
The goal is that data, once committed, is not lost.
Without a durability requirement, systems may acknowledge writes that are only in memory, or keep a single copy on one disk, and lose data at the first failure.
3. How to Measure It (S.M.A.R.T.)
| Metric | Unit | Typical target | How to measure |
|---|---|---|---|
| Annual durability | % probability of keeping each object | e.g., 99.999999999% for object storage | Storage provider design specification |
| Copies of committed data | Count and locations | ≥ 3 copies in separate failure zones | Architecture review |
| Data loss incidents | Count | 0 | Incident records |
| Write acknowledgment | When a write is confirmed | Only after persistence to durable storage | Code and configuration review, crash tests |
Durability figures from storage providers are design targets. They do not cover deletion by users or software bugs, so backups remain necessary.
4. Example Requirements
Each requirement below is ready to copy into a specification: specific, measurable, achievable, relevant and time-bound.
- Transactions: “A payment shall be confirmed to the customer only after it has been written to durable storage in at least two availability zones.”
- Archives: “Archived documents shall be stored with an annual durability of at least 99.999999999% and verified with regular integrity checks.”
- Crash test: “Killing the database process during a write test shall not lose any acknowledged transaction.”
5. Questions to Ask Stakeholders
- Which data must never be lost once confirmed?
- How many copies, and in how many locations, are required?
- When can the system confirm a write to the user?
- Which storage services and durability levels are acceptable?
- How will durability be verified?
- How do backups complement durable storage?
6. Common Pitfalls
- Confusing durability with backup. Durable storage faithfully keeps mistakes and deletions too.
- Acknowledging writes too early. Confirming before data is persisted can lose transactions in a crash.
- Single location. Several disks in one room can be lost together. Separate failure zones.
7. Related NFRs
- Data Integrity — durable data must also stay correct.
- Backup — backups protect against deletion and corruption that durability does not cover.
- Availability — durable data can still be temporarily unavailable.
- Reliability — durability is part of a system’s overall dependability.
8. FAQ
It means that, by design, the expected loss is about one object in 100 billion per year.
In databases, it means a committed transaction survives crashes and power failures.
Yes. Durability protects against hardware failure, not against deletions, bugs or ransomware.
9. Acronyms, Initialisms & Terms
Acronyms are abbreviations pronounced as words (like NASA); initialisms are spelled out letter by letter (like FBI). The classification below follows common pronunciation.
Acronyms
| Abbreviation | Stands for | Definition | Ref. |
|---|---|---|---|
| ACID | Atomicity, Consistency, Isolation, Durability | Four guarantees a database gives so that each transaction is processed completely and reliably. | [4] |
| SMART | Specific, Measurable, Achievable, Relevant, Time-bound | Criteria for writing a goal or requirement so that anyone can check whether it has been met. | [5] |
Initialisms
| Abbreviation | Stands for | Definition | Ref. |
|---|---|---|---|
| FAQ | Frequently asked questions | List of common questions with their answers. | [6] |
| NFR | Non-functional requirement | Requirement that describes how well a system must work (its qualities, such as speed or security) rather than what it does. | [3] |
| S3 | Amazon Simple Storage Service | AWS service for storing files (“objects”) in the cloud. | [7] |
Terms
Plain-language definitions of technical and project-management terms used in this article.
| Term | Plain-language definition | Ref. |
|---|---|---|
| Database | Organized store of data that software can query and update. | [8] |
| Incident | Unplanned interruption or degradation of a service. | [9] |
| “Nines” | Shorthand for availability levels: “three nines” means 99.9%. | [10] |
| Stakeholder | Person or group with an interest in the system: users, sponsors, operators, regulators. | [11] |
| Transaction | Unit of work, such as a payment or an order, that must complete fully or not at all. | [12] |
10. Sources and Further Reading
- [1] Amazon Web Services. “Data protection in Amazon S3.” AWS Documentation. https://docs.aws.amazon.com/AmazonS3/latest/userguide/DataDurability.html (accessed Nov. 18, 2026).
- [2] “Durability (database systems).” Wikipedia. https://en.wikipedia.org/wiki/Durability_(database_systems) (accessed Nov. 18, 2026).
- [3] “Non-functional requirement.” Wikipedia. https://en.wikipedia.org/wiki/Non-functional_requirement (accessed Nov. 18, 2026).
- [4] “ACID.” Wikipedia. https://en.wikipedia.org/wiki/ACID (accessed Nov. 18, 2026).
- [5] “SMART criteria.” Wikipedia. https://en.wikipedia.org/wiki/SMART_criteria (accessed Nov. 19, 2026).
- [6] “FAQ.” Wikipedia. https://en.wikipedia.org/wiki/FAQ (accessed Nov. 19, 2026).
- [7] “Amazon S3.” Wikipedia. https://en.wikipedia.org/wiki/Amazon_S3 (accessed Nov. 19, 2026).
- [8] “Database.” Wikipedia. https://en.wikipedia.org/wiki/Database (accessed Nov. 19, 2026).
- [9] “Incident management.” Wikipedia. https://en.wikipedia.org/wiki/Incident_management (accessed Nov. 19, 2026).
- [10] “High availability.” Wikipedia. https://en.wikipedia.org/wiki/High_availability (accessed Nov. 19, 2026).
- [11] “Project stakeholder.” Wikipedia. https://en.wikipedia.org/wiki/Project_stakeholder (accessed Nov. 19, 2026).
- [12] “Database transaction.” Wikipedia. https://en.wikipedia.org/wiki/Database_transaction (accessed Nov. 19, 2026).
Written by Philippe Huard · LinkedIn · Last updated: November 18, 2026 · Header photo by @tyler_mower; inline photo by @brianwangenheim, both on Unsplash.




