Infrastructure · Security · Engineering
Tell us what you need
Home / Apache CloudStack / Why CloudStack
Scope
Short answers to the questions people actually ask before choosing a platform — including the ones where CloudStack is the wrong choice. Written to be forwarded to a sceptical colleague.
The eight questions
Work it out yourself
Public cloud is priced for elasticity. If your load is steady, you pay that premium on infrastructure that never flexes.
Drag the three sliders to your own numbers. The chart and the verdict update as you move them — and if the arithmetic says renting is cheaper, it will say so.
◄ drag to adjust ►
At your numbers
—
Owned cost includes hardware amortised over four years, plus an allowance for power, space, network and the engineering to run it. It does not include your own staff time if you run it yourselves. If the answer is close, it is not a cost decision — decide on control, residency or exit instead.
You are not on your own
One of the quieter advantages of a platform nobody owns: the people extending it are the people running it. The integration surface grows because operators need it to, not because it appeared on a vendor roadmap.
An Apache-maintained Terraform provider and Go SDK, plus a community Ansible collection. Your estate can be defined in a repository rather than in somebody’s memory.
CloudStack Kubernetes Service for managed clusters on your own hardware, with CSI integration for dynamic persistent volumes from 4.22, and a Cluster API provider for teams who want that route.
Ceph, NFS, iSCSI and StorPool among others, alongside KVM, VMware and XenServer under one control plane. Plugin interfaces are public, so support for new backends arrives without waiting on a vendor roadmap.
Added in 4.21. Orchestrate systems CloudStack does not natively manage, without forking the codebase and without carrying a patch set through every upgrade.
Since 4.20, infrastructure events can drive your own systems directly — provisioning, chargeback, ITSM, whatever needs to know.
The users and dev lists are open, archived and answered by the people who wrote the code. Release announcements and security advisories land there first.
Breadth. Somebody has usually already integrated the thing you need, and the code to prove it is public.
The bit nobody else will. Community plugins cover the common cases; the awkward ones — your ERP, your billing system, a portal your own users log into, a storage array with no maintained driver — are development work. We write those, and we support what we write.
An open ecosystem means you are never waiting on one vendor.It does not mean you are on your own.
The objection worth answering properly
This is the question that stops procurement, and and it deserves an argument rather than reassurance. Here are six.
You can read it. So can your assessors, and so can anyone you hire to look. Nothing about the platform is a matter of trust in a vendor's word.
Vulnerabilities go through the CVE process on the project's timetable, not at a vendor's discretion and not held until a convenient release window.
Telcos, public sector operators and hosting providers run this in production with their own security teams reading the code.
No telemetry, no licence server, no activation. CloudStack runs air-gapped because nothing in it needs to reach the internet.
Not on a vendor's release calendar, and not when a support contract renewal says you may.
ISO 27001 and SOC 2 assess your processes and controls. Neither asks what your hypervisor cost.
In practice, infrastructure is almost never breached through the hypervisor. It is breached through exposed management interfaces, weak or reused credentials, unpatched hosts and misconfiguration. That is an operations problem, and it is exactly the same problem on proprietary platforms.
It is also what we are set up to fix.
When we would tell you not to
We hold partnerships with Nutanix, HP and Dell and we deploy proprietary platforms too. That means when we recommend CloudStack it is a recommendation, not a catalogue — and it means we can say this plainly.
If demand triples for four days a month and sits idle the rest, elasticity is worth its premium. Owning peak capacity you use 12% of the time is the expensive option.
If your architecture leans on managed databases, serverless functions and proprietary queues, the migration cost is not the hypervisor — it is rebuilding all of that.
Running a platform is a discipline, not a project. If there is no team and no appetite to outsource operations, a supported appliance will serve you better than an unmaintained cloud.
If one of those is you, we will say so on the first callrather than the third invoice.
Tell us what you run now and what prompted the review. We will tell you whether CloudStack helps — and if it does not, what would.
An engineer replies within two business days.