Virtual machines and containers have different profiles
Operating system compatibility, device access, isolation, backup and dependencies determine when to use KVM or LXC.
PROXMOX VE + ADD SOLUTION
Proxmox with cluster architecture, storage, backup and operations.
An open platform reduces lock-in but does not eliminate engineering decisions. Add Solution builds Proxmox VE environments for enterprise workloads with availability, networking, storage, protection and the operational lifecycle designed in.
Assessment, architecture, deployment, migration, documentation and ongoing operational support guided by the actual criticality of workloads.
VIRTUALISATION WITHOUT SHORTCUTS
Licensing changes. Responsibility for the architecture remains.
Proxmox Virtual Environment brings together KVM virtualisation, Linux containers with LXC, web-based management, clustering, networking and integration with different storage options. This convergence simplifies the management layer, but availability and performance still depend on well-designed quorum, networking, storage, capacity and operational procedures.
A cluster is not synonymous with continuity
Quorum, fencing, accessible storage, remaining capacity and application behaviour must be validated together.
Updates and recovery are part of the design
Runbooks, maintenance, monitoring, backup and restore tests prevent the platform from relying on informal knowledge.
The project starts with the workload and operational risk—not with hypervisor installation.
CAMADAS DO AMBIENTE
Clusters, networking, storage and protection must evolve as one system.
Add Solution separates functions and failure domains before selecting components. The plan considers management, cluster, migration, storage and backup traffic; latency and redundancy; capacity during failures; and the physical limits of hosts, switches and disks.
- COMPUTE
CPU, memory, NUMA, oversubscription, compatibility and reserves for maintenance or failure.
- CLUSTER
Quorum, Corosync, failure domains, HA policies and intervention procedures.
- REDE
Bridges, VLANs, bonds, MTU, redundancy and logical separation according to flows and risk.
- STORAGE
Local, shared or distributed according to latency, consistency, failure behaviour and operations.
- PROTEÇÃO
Backup, retention, deletion protection, identity segregation, external or offline copies and tested restores, according to the architecture.
PROXMOX VE
Integrated features, explicit choices.
Proxmox VE provides management of KVM VMs and LXC containers, clustering, high availability, live migration, firewall, software-defined networking, APIs and multiple storage backends. A feature’s presence does not establish its suitability: compatibility, topology and supported versions must be confirmed in the project context.
Virtual machines with hardware isolation
Suitable when the workload requires its own kernel, different operating systems or full-VM isolation.
Linux system containers
They share the host kernel and require a specific assessment of compatibility, isolation and the application support model.
Restart orchestration after failure
It depends on healthy quorum, storage access, fencing and available capacity. It does not replace application resilience.
API and repeatable infrastructure
APIs and automation tools help standardise provisioning, inventory and controlled changes.
THE MOST SENSITIVE DECISION
Storage is not a minor cluster detail.
Proxmox VE can work with local and shared storage, including ZFS, NFS, iSCSI and Ceph, among other documented options. Ceph enables integrated distributed storage but requires its own design: networking, node count and profile, disks, controllers, failure domains, replication, recovery and usable capacity cannot be inferred from raw capacity alone.
Latency and the I/O path
The workload must be characterised by IOPS, throughput, block sizes, reads/writes, synchronisation and peaks—not just terabytes.
Capacity during failure and recovery
The design must support rebalancing and recovery without assuming that all nominal resources will be available.
ZFS and local storage
They may be suitable in specific topologies, provided that redundancy, memory, disks, backup and failure impact are addressed.
Ceph is a distributed architecture, not a checkbox for “high availability”.
PROTECTION BEYOND SNAPSHOTS
Snapshots support operations. Backup supports recovery.
Proxmox Backup Server integrates incremental backup, deduplication, compression, verification and retention policies for Proxmox environments. Protection design must also reduce the risk of improper deletion: segregated credentials, least privilege, a destination outside the failure domain and, depending on the architecture, external or offline copies, or copies to a destination with appropriate controls. These measures do not automatically equate to WORM or native immutability. Restore tests must demonstrate that the RPOs and RTOs defined by the business are met.
- CLASSIFICAR
Criticality, RPO, RTO and dependencies per service.
- PROTEGER
Policy, frequency, retention and workload consistency.
- SEPARAR
Independent destination, identity and failure domain.
- VERIFICAR
Integrity, jobs, capacity and anomalies monitored.
- RESTAURAR
Documented file, VM, application and environment tests.
CONTROLLED TRANSITION
Migrating a hypervisor also means migrating dependencies and procedures.
The plan identifies operating systems, disks, drivers, networks, integrations, maintenance windows, application licences and conversion mechanisms. Workloads are grouped by complexity; pilots validate performance and operations; each wave has acceptance and rollback criteria.
- DESCOBRIR
Technical inventory, dependencies and criticality.
- CLASSIFICAR
Compatibility, method and risk for each workload.
- PILOTAR
Conversion, networking, storage, backup and operations.
- MIGRAR
Waves with defined maintenance windows, acceptance criteria and rollback.
- ESTABILIZAR
Monitoring, documentation and operational handover.
CICLO DE VIDA
The platform must remain predictable after go-live.
Operations cover cluster and quorum health, compute and storage capacity, disk failures, latency, backup jobs, certificates, privileged accounts, vulnerabilities, compatibility and updates. Changes are validated against the support matrix and carried out with a backup, a maintenance window and a rollback plan.
Signals that anticipate downtime
Actionable alerts for quorum, memory pressure, latency, capacity usage, media errors, backup and VM status.
Changes with compatibility and rollback
Repositories, subscriptions, release notes, node sequence and testing must follow a defined procedure.
Privileges and auditing
Roles, authentication, service accounts and operational segregation reduce exposure and human error.
Growth and failure headroom
CPU, memory, I/O, network and retention trends guide expansion before operational limits are reached.
OPEN SOURCE COM SUPORTE DEFINIDO
Software use, subscription and support are separate decisions.
Proxmox VE is distributed under an open-source licence and can be used without a subscription. Commercial subscriptions are priced per occupied CPU socket and provide access to the Enterprise Repository and support levels according to the current plan. Repositories, eligibility, response times and scope must be confirmed directly in Proxmox’s current pricing table and terms before purchase.
Add Solution support covers the technical and operational scope agreed with the customer. A Proxmox subscription, when purchased, provides access to the Enterprise Repository and manufacturer support according to the current plan and terms; one service does not automatically replace the other.
PERGUNTAS FREQUENTES
Decisions that need answers before the project starts.
Is Proxmox suitable for critical environments?
It can be, provided the workload, hardware, quorum, network, storage, backup, support and operations meet the service’s availability and recovery requirements.
How many nodes does the cluster need?
For reliable HA, Proxmox documentation recommends at least three nodes, allowing a majority when one node becomes unavailable. In specific topologies with an even number of nodes, an external QDevice can provide an additional vote, provided latency, independence and network failures are assessed. This does not make three nodes a universal answer: remaining capacity, failure domains, maintenance and storage still guide the design.
Is Ceph mandatory?
No. Proxmox supports different storage options. Ceph is a distributed storage option when scale, networking, disks, operations and the I/O profile justify its complexity.
Does LXC replace Kubernetes or application containers?
Not directly. LXC in Proxmox provides Linux system containers; its operational model and isolation differ from those of an application container orchestration platform.
Does a snapshot replace a backup?
No. A snapshot normally remains tied to the same environment and failure domain. Backup requires a policy, retention, separation and tested restores.
Can workloads be migrated from another hypervisor?
Yes, in many scenarios, but the method, downtime and adjustments depend on the disk format, operating system, drivers, networking, applications and tools available in the chosen version.
SOLUÇÕES RELACIONADAS
Tecnologia conectada à operação.
ADD SOLUTION + PROXMOX
Assess the platform in the real context of your workloads.
Talk to Add Solution to map workloads, availability, storage, networking, backup and operations before defining the architecture.