Cloud-B

Services

One account, one API, one private network. Technical specifications are below; prices and plans are on a separate page.

compute

B-Compute

Virtual machines with local NVMe. Disks are not network-attached: write latency is stable and IOPS are not shared with neighbours on the hypervisor.

ParameterValue
HypervisorKVM, QEMU 9.1, host kernel 6.12 LTS
CPUup to 32 vCPU, host-passthrough, AVX2/AVX-512 where available
Memory1–256 GiB, 2 MiB hugepages on request
Diskslocal NVMe, 40–2000 GB, TRIM, snapshots without downtime
ImagesDebian 12/13, Ubuntu 22.04/24.04, AlmaLinux 9, Rocky 9, custom qcow2
Networkup to 4 interfaces per instance, public IPv4 + /64 IPv6
Migrationlive migration within a zone; cross-zone only through a snapshot
Metadatacloud-init, config drive, SSH keys through the API

No vCPU oversubscription is used. If a zone runs out of free cores, the API returns capacity_exhausted instead of creating an instance with reduced limits.

s3

B-Storage

Object storage with an S3 API. Suitable for backups, media and static assets; works with aws-cli, boto3, rclone and any S3 client.

ParameterValue
APIS3, AWS Signature V4; path-style and virtual-hosted-style
Replication3 copies within the zone, asynchronous to geo backup
Consistencyread-after-write for new objects, eventual for list
Versioningenabled by default, previous versions kept for 30 days
Classeshot — immediate access; cold — restore within 12 hours
EncryptionSSE-S3 (AES-256-GCM), one key per bucket
Limitsobject up to 5 TB (multipart), 1000 buckets per account
Compatibilitypresigned URLs, lifecycle, bucket policies, CORS

postgres

B-DB

Managed PostgreSQL and MySQL with automatic failover. A cluster is created in the chosen zone; replicas can be placed in the neighbouring zone.

ParameterValue
EnginesPostgreSQL 16.4, MySQL 8.4 LTS
Topology1 primary + up to 3 read-only replicas
Failoverautomatic, 30–60 seconds, no loss of committed transactions
PITR7 days, 5-minute granularity (WAL/binlog in object storage)
Backupsdaily full, retained for 14 days
Accessprivate network only, TLS required, pgBouncer for PostgreSQL
Extensionspg_stat_statements, pgcrypto, PostGIS, pg_trgm on request

vpc

B-Net

Private networks that do not reach the internet without an explicit rule. Public addresses are issued separately and are not tied to an instance forever.

ParameterValue
VPCup to 16 subnets per network, masks /24 … /29
Security groupsstateful, up to 64 rules per group, dropped-packet logging
Public addressesfloating IPv4, IPv6 /128, movable between instances
Routingstatic routes, NAT gateway, private link between VPCs
MTU1500 by default, jumbo 9000 inside a VPC on request
ProtectionL3/L4 filtering at the edge, 3 Tbit/s capacity per site

Inbound traffic to 25, 465, 587 and 3306 is blocked at the edge by default. These ports require a support request describing the use case.

dns

B-DNS

Authoritative zones managed through the API. Secondary servers in two regions, and zone changes are applied atomically.

ParameterValue
Record typesA, AAAA, CNAME, MX, TXT, SRV, CAA, NS, ALIAS
TTL30–86,400 seconds
Resilience4 servers in 2 regions, anycast announcement
DNSSECsupported, scheduled KSK/ZSK rotation
APIzones and records, batch changes, 30-day version history
Limits100 zones, 5,000 records per account

backup

B-Backup

Scheduled snapshots of volumes, buckets and database clusters. Copies are moved to the other region and can be restored through the API or with support.

ParameterValue
SourcesVM volumes, B-Storage buckets, B-DB clusters
Schedulecron, up to 24 points per day per resource
Retention7–90 days, geo backup in the other region
Restorevolume — up to 15 minutes (RTO), object — on request
Verificationmonthly restore test on the provider side
Encryptioncopies encrypted with a separate account key

Included in every plan

Basics are not sold separately.

IPv6 and private networking

A /64 per instance and private networks at no extra cost.

API and Terraform

A complete set of operations, idempotent requests, a provider in the registry.

Status and postmortems

A public status page and an incident write-up within five business days.

Snapshots through the API

Without stopping the instance, stored in the geo backup.

Host monitoring

Hypervisor and network metrics available in the API and a Prometheus endpoint.

24/7 support

Email and phone, on-call engineers, a 30-minute response for P1.