System Architecture

Architecture diagram

Availability

The service is constructed on the AWS cloud. Including the VSS engine (database), every service component is operated in a high-availability cluster.

The service has a scale-out architecture with multiple versioned copies of the vector database. To ensure data integrity, all write operations are performed in two phases. In phase one, the write operation is assigned a serial number is committed to a write-ahead-log file. The write-ahead-log file stores all the pending write operations in an ordered format. The ordered format ensures that all write operations are applied to our cluster nodes in the same order to avoid corruption due to synchronization issues. The database stored on each node is also versioned, and the serial number tracking the latest write operation is stored to ensure consistency and resilience in file corruption. Checkpoints occur every 24 hours to ensure that we can always restore the database if a catastrophic failure loses at most 24 hours of data.

Our target uptime is 99.99% (not strictly defined as an SLA). We do not offer any compensation even if this target is not reached.

Please visit our Status Page to check the service status.

Service regions

JCV Cloud AnySee is a global cloud service that operates in multiple service regions, helping international firms to use a distributed data structure to meet local legal restrictions. There are no differences in the API function between different regions. Data are isolated between each service region, and we do not yet offer data migration services.

As of April 2023, AnySee operates in the following service regions.

• 🇯🇵 ap-northeast-1
• 🇸🇬 ap-southeast-1
• 🇺🇸 us-east-1
• 🇧🇭 me-south-1