Skip to content

Services · Cloud Compute

Cloud compute built for production.

VMs, bare metal, private VLANs, and all-flash distributed storage, run by senior engineers in two Montréal data centres. For workloads someone has to keep running on Monday morning.

Request infrastructure plan Compare with Cloud PaaS

Client portal

Inside the Stack Harbor client portal.

Screens from the portal, shown with sample data.

The Cloud Servers list in the Stack Harbor client portal
Every cloud server in one list. Start, stop, reboot or rename any of them. How it works →

What you can deploy

Compute capabilities.

  • Virtual machines

    Configurable compute, memory, storage, and network. Provisioned in minutes, resized live.

  • Bare-metal & dedicated

    Single-tenant hardware for workloads that need consistent performance or compliance isolation.

  • Private networking

    Private VLANs, IPv4/IPv6, and secured internal traffic.

  • All-flash SSD storage

    High IOPS, distributed, block and object, with capacity tiers tuned to the workload.

  • Hypervisor resilience

    Live migration off failing hosts, regional pairing for managed cutover. Application-level HA lives on Cloud PaaS.

  • Managed operations

    Architecture, monitoring, updates, support, migration, and technical escalation: your environment, our team.

VM shapes

Pick a shape, resize it live.

Cloud Compute provisions VMs from a small set of shapes that cover most production workloads. Every shape supports live vertical resize without rebuild. Custom shapes available when the workload requires them. Workloads that need managed load balancing and automated horizontal scaling belong on Cloud PaaS.

ShapevCPUsRAMSSDNetworkBest for
Compact
s1
 2 4 GB 80 GB 1 Gbps Dev, staging, utility, jump host
Standard
s2
 4 8 GB 160 GB 1 Gbps General-purpose app servers, WordPress, mid-tier APIs
Performance
p1
 8 32 GB 400 GB 10 Gbps High-throughput web, mid databases, WooCommerce checkout
Memory-optimized
m1
 8 64 GB 400 GB 10 Gbps Postgres / MariaDB primary, Redis, Elasticsearch

Vertical resize is live and reversible. Horizontal scale-out is a planned operation, not automated — when the workload needs automated scale and managed load balancing, Cloud PaaS is the fit.

Use cases

When teams pick Cloud Compute.

  • High-traffic marketing and publisher sites
  • cPanel/WHM hosting platforms
  • Agency client portfolios
  • SaaS application backends
  • WordPress at scale
  • Private cloud environments
  • Migration landing zones
  • Custom production workloads
  • Compliance-constrained deployments

Reference topology

A production IaaS environment looks like this.

A typical Stack Harbor IaaS deployment runs in our two Montréal data centres with private VLANs, off-site backup, and a managed cutover playbook. Application-level HA (auto-scaling, managed load balancing) lives on Cloud PaaS — the diagram below shows the IaaS substrate.

Internet · users
Edge
CDN · DNS · WAF · TLS
Montréal · DC 1
PRIMARY
  • LB Load balancer · optional
  • App tier Compute pool · vertical resize
  • Cache Object cache · Redis
  • Storage Distributed SSD · block + object
Montréal · DC 2
SECOND SITE
  • LB Load balancer · optional
  • App tier Compute pool · vertical resize
  • Cache Object cache · Redis
  • Storage Distributed SSD · block + object
Backup
Backups included; ask us for the schedule
live traffic data replication managed cutover

VM management

Every VM, the same control surface.

From the dashboard, every Cloud Compute VM looks the same: an Overview pane with state and resources, plus tabs for Networking, Storage, Backups, and an in-browser Console. Toggle auto-backups or autoscale per VM; reboot, reinstall, or migrate from a single screen.

Self-serve in the customer portal · scriptable through the public API.

What happens when…

Operations, not surprises.

The value of managed compute shows up the moment something happens. Here is the concrete response Stack Harbor runs for the operational moments that show up in every production environment.

  • Workload doubles overnight

    Live resize

    Vertical resize live — more vCPUs and RAM, no rebuild, no IP change. Horizontal scale-out is a planned operation we coordinate with you; workloads that need automated scaling under load belong on Cloud PaaS.

  • Region degraded

    DR runbook

    Recovery follows your DR design: cutover where the design includes a standby, otherwise restore from off-site backup. Run by Stack Harbor on alert, with your sign-off on the call. Postmortem after.

  • Need a point-in-time rollback

    Snapshot · restore

    Instant snapshot, off-host backup at the configured RPO. Restore is a portal action or an API call — minutes, not hours, for a typical VM.

  • Disk approaches capacity

    Online extend

    Alert fires at the threshold you set. Online LVM extend on the same VM, no reboot, no downtime. Add a volume if a new mount point makes more sense.

  • Hardware fails under the VM

    Live migration

    Live migration to a healthy host on the same fabric, transparent to running connections where the workload allows. We open the postmortem.

  • Compliance asks for audit evidence

    Audit log

    Activity log exported with timestamps and actor per action — VM lifecycle, backup runs, network changes, console sessions. Maps cleanly to SOC 2 / ISO 27001 evidence packs.

These are the operational defaults — coverage windows, response SLAs, and postmortem inclusion scope to your plan.

Operational substance

What the infrastructure does.

  • High IOPS storage

    All-flash SSD distributed storage tuned for production read/write.

  • Hypervisor resilience

    Live migration off failing hosts; regional cutover is a managed operation, not automatic. Application HA belongs on Cloud PaaS.

  • Live monitoring

    Application, network, resource, and synthetic checks.

  • Backup policy

    Backups included; ask us for the schedule.

Pricing model

Compute pricing maps to workload, not packaged tiers.

Cloud Compute is priced as usage-based resources, managed monthly plans, or a custom managed environment, depending on workload profile, traffic, storage, support, SLA, and management scope.

Pricing drivers

  • Compute and memory profile
  • Storage tier (SSD / capacity / mixed)
  • Public and private network needs
  • Support and SLA tier
  • Managed scope (DIY · partial · fully managed)
Request infrastructure plan

Where Cloud Compute lives in our offer

The solutions that buy it.

Cloud Compute is rarely bought on its own. It's the substrate behind the engagements below.

From the blog

Guides for the decision you are making

All articles →

Need compute someone is accountable for?

Talk to a Stack Harbor engineer. We scope the environment around your workload.

Request infrastructure plan Compare with PaaS