Network Volumes
Network volumes are persistent block devices suitable for scaling volume space without changing the basic volume. Three-fold replication of the volumes ensures high data safety.
A network volume can be created together with the server or, unlike the local volume, separately from the server.
Network volumes have the following features:
- can be both a boot (system) volume of the server and can be connected as an additional volume to expand volume space;
- up to 255 network volumes can be connected to the server when using the standard virtio-scsi (when using ide – up to 4, when using virtio-blk – up to 26);
- can be detached from the server;
- easy to transfer between servers in the cloud — detach from one server and connect to another;
- 20 TB is the recommended limit for the network volume;
- you can create a snapshot of the network volume.
Network Volume Types
Network volume types:
- Basic volume — based on enterprise-class SATA volumes. Best for storing large amounts of data with few read and write operations;
- Universal volume — based on SSD. Best for storing system volumes of virtual machines;
- Fast volume — based on SSD. Faster response time and higher performance compared to basic volumes. Best for workloads that require increased I/O.
The bandwidth settings and read/write limits in the IOPS of network volumes depend on the pool. Learn more in the Scalable Network Volumes section on our website and in the table.
Network Volume Limits
IOPS is the number of read and write operations on the volume per second. The higher the volume performance, the faster these operations are completed.
When you first launch the cloud server, the file system on the system volume “stretches” to fit the volume. The larger the volume size and the lower its IOPS limits are, the more this process will take, and therefore, the longer the cloud server will start.
The file system size also affects the time it takes to check the status of the file system in the event of a server crash. This check is enabled by default for the system volume in all servers created from the Cloud platform’s prepared images and can be manually configured for connected volumes.
Depending on the selected volume type, purpose of use, and the possibility of further fine-tuning of the file system, we recommend that you do not exceed the limits indicated in the following table. It also contains the IOPS limits.
| Volume type | Recommended limits for the system volume | Recommended limits for the additional volume | Bandwidth | Read/write volume limits |
|---|---|---|---|---|
| Basic (all pools) |
500 GB | 10 TB | 100 MiB/s | 320 IOPS / 120 IOPS |
| Universal (ru-1, ru-2a, ru-2b, ru-3) |
2 TB | 10 TB | 150 MiB/s | 640 IOPS / 320 IOPS |
| Universal (ru-2c, ru-7, ru-8, ru-9) |
5 TB | 10 TB | 200 MiB/s | 7000 IOPS / 4000 IOPS |
| Fast (ru-1, ru-2a, ru-2b, ru-3a, ru-3b) |
2 TB | 10 TB | 500 MiB/s | 12800 IOPS / 6400 IOPS |
| Fast (ru-2c, ru-7, ru-8, ru-9) |
10 TB | 10 TB | 500 MiB/s | 25000 IOPS / 15000 IOPS |
The data is provided for volumes with GPT partitioning, as this partitioning is used by default in the Cloud Platform’s prepared images.
Creating Your Own Image with GPT Partitioning Format
If you prepare images yourself and plan to create volumes with a capacity of more than 2 TB from these images, use GPT partitioning instead of the common MBR partitioning. Learn more about partitioning formats in the article GUID Partition Table.
The procedure for creating your own image with GPT partitioning format is described in the article.