Our Cloud Servers are provisioned on fully redundant, distributed infrastructure. The infrastructure offers n+1 redundancy and was designed with high availability in mind.
Cloud Nodes (Compute)¶
Cloud Nodes are based on enterprise server hardware, using Intel Xeon and AMD EPYC CPUs, ECC Memory as well as 10Gbit network cards. KVM as well as Corosync are utilized to run virtual machines (Cloud Servers) and provide automated failover in case of node failure.
Load is shared between Cloud Nodes by live migrations, depending on the current cluster load.
Storage is provided by n+1 redundant, distributed+replicated as well as arbitrated GlusterFS Volumes to provide high availability as well as increased performance. Our storage environment behaves similar to Raid 10, beside the fact, that every node is already a server with Raid 10 + NVMe Cache, hotswappable SSD/HDD/NVMe Disks, Dual 10G network connectivity as well as Dual PSU.
Storage Nodes and Cloud Nodes are interconnected over a physically seperated, n+1 redundant Juniper QFX based 10G/40G storage network.
Daily Backups are done to single backup nodes, connected with 10Gbit/s or more. Every Backup Node is equipped with Raid 10 and Dual PSU. Backups are created by a self developed backup solution called “Cloud Backup”.