Enhancements on Virtual Machine
Special for Edge DC
Management of Diversified Resources
TECS Cloud OS supports the unified life cycle management of VMs, bare metal, and acceleration resources to meet diversified resource requirements of customers.
Performance enhancement: The vCPUs of the VM can be bound with the CPUs of the physical host, and the vCPUs and memory of the VM can be automatically allocated from the same NUMA node. In addition, the memory of the VM supports huge page to improve the operation performance of the VM. The VM supports SR-IOV and DVS distributed routers as well as multiqueue technology, improving network performance of the VM. SR-IOV and DVS can share the same physical NIC for flexible configuration. It also supports RDM function which allows the VM being directly mounted with raw disks to improve the I/O performance, and supports the real-time operating system and real-time VM to improve the real-time performance.
Scheduling strategy: The VMs can be deployed in host or host aggregates as the unit to deploy various kinds of VMs, such as affinity group (VMs in the group are deployed on the same physical node), anti-affinity group (VMs in the group are deployed on different physical nodes), soft affinity group (Affinity group in preference. If resources are not enough, VMs can be deployed freely), and soft anti-affinity group (Anti-affinity group in preference. If resources are enough, VMs can be deployed at random), VM e occupying CPUs exclusively, VM occupying NUMAs exclusively, VM assigning nodes/NUMAs, and centralized/decentralized deployment of VMs, meeting customers’ different requirements.
Dynamic adjustment: The system supports hot-swappable disks, NICs, and online/offline migration of storage. Meanwhile, it supports the change of specifications/hard disks of VMs to meet real-time resource requirements of service VMs.
QoS function: By flexibly controlling the quality of the CPU, IO and network bandwidth of VMs, it provides controllable computing capability of VMs, so as to ensure or restrict the VM’s capability to use computing resources within the specified range. In this way, the VMs with different service requirements can obtain the most appropriate computing performance, thus achieving optimal resource reuse and cost reduction.
2. Bare Metal
Lifecycle management: It supports the query, registration, update, discovery, recovery and deletion of bare metal to implement convenient management of bare metal and meet the customer's requirements.
Co-management of VM and bare metal: The solution can simultaneously manage VMs and bare metal in a system to address diversified application deployment scenarios.
Management enhancement: The solution supports to divide bare metal into AZs to meet the requirements of bare metal partition management, and supports bare metal KPI to monitor the health status of bare metal in real time.
3. Accelerated resources
Life cycle management: The solution supports automatic discovery, query, enabling, application and allocation of hardware acceleration resources, so that acceleration hardware can be used as cloud resource on demand.
Management enhancement: The solution supports unified management of acceleration resources used by VMs and bare metal. It also supports such O&M management functions as acceleration resource KPI and status monitoring to monitor the health status of acceleration resource.