One object, one cluster
Versions of one file are never spread across regions. Otherwise cleanup of unused objects, deduplication and quotas would turn into a distributed problem with no gain.
Data residency
A3.Drive keeps the original of every file in the cluster of the region assigned to the account. This page explains how that works in the storage. The general country requirements for personal data are collected on a separate page for all A3 products.
Region is a property of the object, not of the request
The storage distributes originals by region: an object physically sits in a cluster, that is a group of storage servers in one country. This is not a content delivery network. A delivery network is a read cache, a copy closer to the reader. A region is where the original sits: it decides where writes go and which jurisdiction the data is in. One does not replace the other, and they work together.
- The file identifier is global and does not contain the region, so a move between clusters does not break bindings
- The storage location is described by fields of the object: the region and the coordinate inside the cluster
- The region belongs to the account, not to the request IP address: an employee travelling does not move the data
- Backups are kept in the same region as the original
Countries of the A3 line
A reference for choosing a storage region. It is not a legal opinion: requirements are confirmed by lawyers in each jurisdiction, and in this area they change often.
CountryPersonal data localizationStorage region
KazakhstanStorage on the territory of the republicRegional cluster
RussiaPrimary collection and storage of citizens data in databases on national territorySeparate cluster needed
UzbekistanProcessing of citizens data using databases on national territorySeparate cluster needed
KyrgyzstanA strict localization requirement is unconfirmed, cross border transfer is regulatedBeing confirmed with lawyers
MongoliaA data protection law is in force, limits concern cross border transferBeing confirmed with lawyers
Rules built into the data model
Adding a region to a system built without one means rewriting the data access layer in every product at once. That is why the region has been part of the storage core from day one, however many countries are served.
Across regions a cache works, not a copy
A file from one country shared with a partner in another is read through a cache over the original. The original is not copied into a second cluster and stays in the owner region.
Deduplication within a region
Deduplication keeps identical files as one object. Two customers in different clusters with a byte identical file get two objects: looking for duplicates across a region boundary would break data sovereignty.
The link leads to the right cluster
When a file is requested, the storage looks at the object region and issues a signed link to its cluster. Products know only the file identifier and keep no addresses.
Country requirements for data
This page covers only how the region works in the storage. Personal data laws, cross border transfer and the choice of a hosting country for all ecosystem products are described on a shared hub page.
Shall we discuss placement for your jurisdiction?
Tell us which countries the company operates in and whose personal data is processed. We will answer which storage region fits and what it takes.