The release focuses on three key areas: automatic load balancing between compute nodes, disk space savings through linked clones, and support for modern storage protocols. In total, the new version includes over 60 improvements and enhancements.
Dynamic Load Balancing
Basis Dynamix Enterprise 4.6 introduces a Distributed Resource Scheduler (DRS) mechanism, which ensures interaction between the platform and the Basis Resource Optimizer module. Together, they balance the load on compute nodes — monitoring the infrastructure's state and, if necessary, redistributing virtual machines between nodes for more even cluster resource utilization, which helps reduce equipment downtime.
Central panel of the Basis Resource Optimizer cluster
Basis Resource Optimizer analyzes CPU and memory consumption metrics and, based on them, generates recommendations: for migrating virtual machines between nodes for even load distribution, for changing the number of vCPUs and RAM, and also monitors compliance with placement rules — Affinity (keeping VMs on one node) and Anti-Affinity (distributing them across different nodes). Depending on the selected mode and rule settings, recommendations are applied automatically or after operator confirmation.
Operations that require a virtual machine reboot, such as changing its allocated resources, can be performed at predefined intervals. This is handled by the technical windows service: the administrator sets one-time or recurring schedules, and such tasks are planned for the nearest allowed window during periods of minimal load.
Distributed Resource Scheduler Basis Dynamix Enterprise
DRS within Basis Dynamix Enterprise is installed in the management cluster. The administrator can create schedulers within individual zones, add and remove nodes from their scope, and start and stop the mechanism via API. This allows customer administrators to configure balancing policies to suit the specifics of a particular infrastructure — from compact clusters to large geo-distributed installations.
Linked Clones and Disk Space Savings
For data storage on local disks of compute nodes (Local SEP), the platform now supports thin disks and linked clones. If a virtual machine is created from an image located in the same pool, its disk becomes a dependent descendant of that image — actual data copying is not performed, and changes are written using the copy-on-write principle. This reduces disk space consumption during mass deployment of identical VMs — for example, in VDI scenarios, on testbeds, or in service provider installations.
In parallel, support for the “MultiImage” mechanism has been introduced, which combines links to copies of a single template image located in different storage pools. This allows creating linked clones from a single source on various storage systems and reduces the dependence of deployment operations on the state of an individual pool.
Virtual machine disks available in Basis Dynamix Enterprise
In the administrative portal for pools with thin disks, physical limit and actually written volume parameters are displayed with color-coded fill levels: the pool is highlighted in yellow when 75% full and red when 90% or more — the administrator immediately sees where intervention is required.
NVMe-over-TCP Support
The platform has gained support for the NVMe-over-TCP protocol for storage systems connected via SEP SHARED and SEP Tatlin in RDM mode. This network protocol provides access to block devices with performance close to local NVMe, using standard Ethernet infrastructure — without the need to deploy a separate Fibre Channel fabric or specialized RDMA equipment.
NVMe-over-TCP support expands Basis Dynamix Enterprise's compatibility with modern storage systems and gives customers the ability to build high-performance storage systems on familiar network infrastructure.
Management via Virtual Controller
The new release expands the set of tools for platform administration and operation. In particular, the ability to administer Basis Dynamix Enterprise using virtual management nodes (controllers) has been added. Thanks to this, users do not need to allocate physical nodes for the platform's management layer if its size does not exceed 40 nodes. It also became possible to back up the management node with classic tools, restore it faster in case of failure by storing the node on the storage system, and move virtual machines between nodes to manage the load.
In addition, Basis Dynamix Enterprise 4.6 features redesigned interfaces for creating virtual machines and managing zones, simplified deployment of new installations, expanded network capabilities, and updated platform service components.
“In the new release, we have taken another significant step from manual administration to automated infrastructure management, regardless of its size and complexity. The DRS resource scheduler relieves the administrator of the routine task of load balancing, linked clones and thin disks reduce disk space consumption, and NVMe-over-TCP support opens access to modern storage architectures without changing the network infrastructure. This is a consistent movement towards a platform that takes on more and more operational tasks and leaves administrators room for service development,” commented Dmitry Sorokin, Technical Director of Basis.