Skip to main content

S3 Migration FAQ

Where is S3 physically located?

S3 is located in different locations across three countries — Russia, Kazakhstan, and Uzbekistan. The location is selected when creating a bucket.

You can view the locations where S3 is available in the S3 subsection of the Product Availability by Location instructions. You can view location addresses in the Selectel Infrastructure subsection of the Locations: Countries, Regions, Availability Zones, Pools, and Data Center Addresses instructions.

Which method should I use to migrate from a third-party S3 now?

We recommend migrating using tools, such as Rclone. For more information, see the Example of migration to S3 via Rclone instructions.

Migration to S3 via the control panel may fail during extremely high load on the service. You can track the status of the migration service on the status page.

How do I use S3 to store infrastructure backups?

You can store backups, images, and other data in S3.

For example, to transfer a disk image to S3, use the Transferring a disk image to a third-party S3-compatible storage instructions.

What affects migration speed?

Selectel does not affect the migration speed on your provider's side. The speed also depends on intermediate network components. We are working to increase the migration speed on our side, but we cannot guarantee faster migration in every case.

To increase migration speed:

  • use the API and tools, such as Rclone; for more information, see the Example of migration to S3 via Rclone instructions;
  • configure migration to multiple buckets rather than a single one;
  • distribute data across multiple locations;
  • increase the number of data threads, for example, when using Rclone.

How is data replicated in S3?

All data in S3 is replicated. The replication method depends on the selected pool:

  • multi-AZ pool (ru-6) — data replicas are synchronously distributed across three physically isolated data centers (DCs). Tolerates a complete failure of one data center without data loss or service downtime;
  • single-AZ pool (ru-1, ru-3, ru-7, gis-1, uz-2, kz-1) — data replicas are synchronously distributed within a single data center across multiple independent physical servers in different racks. Tolerates the failure of individual drives, servers, or racks within a single DC.

Automatic data replication to other pools, automatic migration, or failover to another pool is not supported. You can configure backups to another pool, region, or country manually, or store backups on a third-party platform.

How do I further protect my data in S3?

You can:

Are data backups supported in S3?

S3 supports data replication, but does not create data backups that can be used for recovery in case of unavailability.

You can configure scheduled backups from S3 on your own using third-party tools. For example, you can set up backups to an S3 bucket in another pool, availability zone, or country using:

Why is cold storage unavailable in some locations?

To increase the total physical capacity and performance of S3 for new buckets, we are consolidating storage classes in certain regions. The storage class of existing buckets remains unchanged.

You can view current information on storage classes available for ordering in a specific location in the control panel when creating a bucket under Configuration → Storage class.

What data security guarantees are provided for S3?

The terms of use for the S3 product and the guarantees provided by Selectel are defined in the Terms of Use for Specific Services: "Object Storage S3" and SLA . For more information about security at Selectel, visit selectel.ru and see the Security Guide for Selectel Products section.

You can enhance the security of your data in S3 on your own.

How much data can be transferred?

There are no limits on the amount of data that can be transferred on the S3 side. The storage is scalable.

There are limitations on certain storage entities and the number of API requests. You can set S3 limits and bucket limits on your own — they are disabled by default.

Transferring large amounts of data may take time, particularly during high load on the service.

How do I calculate the cost of migration to S3?

Regardless of the migration method and tool, the ability to migrate to S3 is provided free of charge.

However, in Selectel, the S3 bucket is billed according to the S3 payment model.

If the migration is performed:

  • from third-party S3-compatible storage: API requests and storage usage are billed in the target bucket in Selectel. Inbound traffic is not billed. Outbound traffic is billed on the side of the S3 storage provider from which the migration is performed;
  • from another Selectel S3 bucket: API requests are billed in the source bucket, and API requests and storage usage are billed in the target bucket. Traffic between buckets is internal within Selectel and is not billed.

Can I transfer a bucket to another location?

You can move data to another pool — in the same or another data center (availability zone), in the same or another country.

To transfer data, use available migration methods. After migration, you can delete the data from the source bucket or keep it in multiple locations to improve fault tolerance.

Requests to S3 fail with an error, what should I do?

Currently, high load on S3 is observed in all locations. Because of this, requests may fail with 5xx errors or be interrupted due to timeouts.

We recommend setting longer intervals between retries and distributing data across different S3 locations. Recommendations:

  • do not retry the request immediately — this increases the load and often leads to a series of 409 and 5xx errors;

  • configure retries for requests with 5xx errors and timeouts using an exponential backoff scheme:

    • initial pause — 10 seconds;
    • maximum pause — 60 seconds;
    • sequence — 10 → 20 → 40 → 60 seconds;
    • total duration of attempts — 5 minutes.

  • reduce the amount of concurrently uploaded data;

  • do not exceed API request limits;

  • abort multipart uploads (AbortMultipartUpload) if a multipart upload (Multipart Upload) was interrupted and you do not plan to complete it. Otherwise, it remains incomplete and already uploaded parts will be billed;

  • distribute data across multiple S3 locations so that peak load in one location does not slow down your work. If mass errors occur in one region, switch new uploads to another.

How do I work with S3 after migration?

For more information about product features, operating principles, and ways to work with S3, see the S3 Product overview instructions. To work with storage, buckets, and objects, use the S3 section.

We also recommend taking the course Product Dive: S3 from Selectel on the Selectel.Stack learning platform.