What is IaaS?

A provider's IaaS offering is usually assembled from a handful of parts:

• Compute - virtual machines or dedicated hosts, described by processor count, memory, instance family and storage policy.
• Storage - volumes that outlive the machines they attach to, plus images, snapshots and backups.
• Networking - virtual networks, subnets, firewall rules, routing and addresses reachable from outside.
• Lifecycle controls - build from an image, start, stop, resize, snapshot, destroy, all through an API.
• Access and quota - credentials, project boundaries and per-user ceilings on what may be created.

Sitting under those parts is the control plane that turns an API call into a running machine. It holds the image catalogue, the network definitions, the quota rules and the live state of every instance it has created. Two providers can publish identical processor and memory shapes and still differ sharply in what that layer permits: which images exist, how networks are separated, how far an instance can grow without a rebuild, and how much of the API is exposed at all. Those are the details worth recording in a specification before a workload is migrated, because they decide whether the migration is a configuration change or a rewrite.

Everything above that boundary stays with the tenant: patching the guest system, installing runtimes, deploying the application, protecting the data. That is the honest trade in this model. IaaS does not delete operational work, it relocates it, which is why teams renting infrastructure tend to buy back engineering time rather than headcount.

IaaS is one of three labels commonly applied to what a cloud hands over, the others being platform services and software delivered as a service. This page stays with the first of them. The side-by-side reading of all three - what each takes off your plate, where the responsibility line moves, how the pricing behaves - lives in IaaS, PaaS and SaaS on VirtEngine.

The object a tenant actually works with is the virtual machine: an emulated computer with its own operating system running on shared hardware. Every instance in a listing is a defined shape of that object - a number of processors, an amount of memory, a starting image, a position on the network.

IaaS suits workloads where the shape of the environment is part of the requirement: applications lifted from a datacenter that were never written for a platform service, clusters needing particular accelerators or interconnects, batch jobs that need machines for an afternoon, and deployments whose data must sit in a specific jurisdiction or network.

What buyers compare reaches past the instance list. Transfer and storage policies often move the total cost as much as the hourly rate does; image availability and region placement decide whether a workload can move at all; and everything above the hypervisor becomes the tenant's security surface - a wider responsibility than a managed service hands over.

On VirtEngine, infrastructure appears as a published offering rather than a negotiated contract. A provider lists the instance shapes it will serve with their region, network characteristics and price, and a tenant funds an escrow account when placing the order. The provider can confirm committed budget before handing over a machine, and the tenant is not paying ahead for capacity it has not received.

The order can match at the listed price, through competing bids, or by selecting listings that satisfy stated attributes; either way it becomes a lease binding tenant, provider and escrow. The provider's control plane fulfils the request, records what was consumed and signs those readings. After the dispute window, validated line items are drawn from escrow and anything left over returns to the tenant, with both sides identity-verified before the sequence starts.