Two storage proposals can quote similar prices today and diverge dramatically five years later. Often the reason is not hardware but licensing. Software-defined storage is usually licensed in one of two ways: by capacity, measured in terabytes, or by node, measured in servers. Each model rewards different growth patterns, and the difference becomes significant at petabyte scale and as drive sizes keep increasing.
This article explains how capacity-based and node-based licensing work, how each behaves under common growth scenarios, what to watch for in contracts and how to compare offers fairly over the life of a platform. It is written for architects, IT directors and procurement teams evaluating object storage and other software-defined platforms.
The two models
Capacity-based licensing
You pay according to how much capacity is licensed, typically per terabyte of usable or raw capacity, sometimes with tiers that lower the price per terabyte as volume grows. Adding capacity, whether through more servers or larger drives, increases license cost.
Node-based licensing
You pay per server or node running the software, regardless of how much capacity each node holds. Adding nodes increases license cost; adding larger drives to existing nodes, or buying denser nodes, does not.
Variations
Real contracts mix these elements: capacity licenses with minimums, node licenses with capacity caps per node, subscriptions that bundle support, enterprise agreements with unlimited use within a scope and separate charges for features such as replication or object lock. Read the details carefully.
How each model behaves as drives get bigger
Drive capacities keep increasing. Each generation of high-capacity hard drives stores more data in the same slot, and dense servers hold more drives.
- Under capacity licensing, larger drives do not reduce software cost. If you store twice as much data, you pay roughly twice as much in licenses, though tiered pricing may soften this.
- Under node licensing, larger drives and denser servers lower software cost per terabyte, because the same number of licensed nodes holds more data.
For organizations whose data grows fast, node-based licensing combined with dense hardware can significantly reduce cost per terabyte over time. For organizations with modest capacity and many small nodes, capacity licensing may be more economical.
Raw, usable or logical?
Capacity licenses can be measured in different ways, and the difference matters:
- Raw capacity counts every byte of disk, including protection overhead.
- Usable capacity counts capacity after protection such as erasure coding or replication.
- Logical or stored capacity counts data stored, sometimes before or after compression and deduplication.
A license on raw capacity effectively charges for protection overhead. A license on logical data before reduction may charge for data you never physically store. Make sure every quote states clearly what is measured, and compare like with like.
Growth scenarios
Steady, fast growth
An organization growing 30 to 40 percent a year for several years, such as a backup, archive or media environment, adds a lot of capacity. Node-based licensing with dense nodes usually wins here, because each new generation of drives adds capacity without proportional license cost.
Modest growth, small footprint
An environment of a few hundred terabytes with slow growth may find capacity licensing simpler and cheaper, especially if it would otherwise need several licensed nodes for availability.
Mixed workloads
A platform hosting backups, archives and application data may grow unevenly. Capacity licensing with volume tiers can work well if tiers are generous; node licensing works well if nodes can be filled densely.
Multi-site deployments
Replication across sites multiplies capacity. If licenses count every copy, capacity-based costs grow with each site. Check whether replicated or erasure-coded copies across sites are counted once or multiple times.
An illustrative comparison
Consider an environment starting at 2 PB usable and growing 30 percent a year for six years, ending above 9 PB. Under capacity licensing, license cost grows roughly in line with that capacity, softened only by volume tiers. Under node licensing, the organization starts with a set of nodes and refreshes with denser servers and larger drives over time. If drive capacity per node doubles over the period, the number of licensed nodes needed to reach 9 PB grows far more slowly than capacity itself. The exact crossover depends on prices, but the shape is consistent: the faster capacity grows and the denser the hardware, the more node-based licensing favors the buyer.
Other contract terms that change the answer
- Support pricing: is support a percentage of license value, a per-node fee or bundled in a subscription?
- Price escalators: annual increases compound over five to seven years.
- Hardware refresh: can licenses transfer to new hardware without repurchase?
- Feature licensing: are replication, object lock, encryption or multi-tenancy extra?
- Minimum commitments and true-up terms.
- Term and termination: what happens at renewal, and what does it cost to exit?
Questions to ask every vendor
- Is the license based on raw, usable or logical capacity, or on nodes?
- Are replicated copies at other sites counted separately?
- Does data reduction change the licensed amount?
- What happens to licenses when hardware is replaced or nodes are retired?
- Are replication, immutability, encryption and multi-tenancy included?
- How is support priced, and how does it change over time?
- What annual price increases apply, and are they capped?
- What are the true-up and minimum commitment rules?
- What does it cost to exit at the end of the term?
Ask for written answers and model each response in the same spreadsheet.
Licensing and backup workloads
Backup and archive environments deserve special attention because they grow continuously and rarely shrink. Longer retention, immutability periods that prevent early deletion and new data sources all push capacity up. A licensing model that looks reasonable at today's capacity can become a major cost line within a few years. Model backup growth explicitly, including immutable retention, and favor terms that keep the cost per terabyte falling as capacity grows.
Licensing for service providers
Service providers selling storage to tenants have different priorities. They need licensing that scales with revenue, allows capacity to be added quickly when a large tenant signs and does not penalize them for protection overhead or replication. Some prefer consumption-based or pay-as-you-grow terms that align software cost with billed capacity.
How to compare offers fairly
- Define the same scenario for every vendor: starting capacity, growth, sites, protection and data reduction assumptions.
- Model five to seven years, including expansions and at least one hardware refresh.
- Normalize to usable capacity and cost per usable terabyte per year.
- Include support, escalators and feature fees.
- Include hardware, since licensing interacts with server density and drive choice.
- Run sensitivity analysis on growth rate and drive size, since both are uncertain.
Licensing and hardware freedom
Software-defined storage that runs on standard servers lets you take advantage of each new drive generation. Licensing that rewards density, combined with freedom to choose hardware vendors, compounds the savings: you buy the densest economical servers available at each expansion, and software cost per terabyte falls. Appliance-based platforms tie software and hardware together, which can limit those gains.
How Scality approaches licensing
Scality offers licensing designed for large and growing environments, and its software runs on standard x86 servers from many vendors, so customers can adopt denser hardware as it becomes available and refresh hardware without migrating data. Scality teams can model licensing against your growth scenario so you can compare offers on a consistent, multi-year basis.
Revisit licensing at every expansion
Licensing is not a one-time decision. At each expansion or renewal, check whether your growth and hardware choices still match the model you chose.
Putting it together
Capacity-based licensing charges for every terabyte you add; node-based licensing charges for every server you add. As drive sizes increase and data grows quickly, node-based models combined with dense hardware usually cost less at scale, while capacity-based models can suit smaller or slower-growing environments. The deciding factors are growth, hardware density, how capacity is measured and the contract terms around support, features and refresh. Compare offers over five to seven years, using the same scenario for every vendor.
Frequently asked questions
What is capacity-based storage licensing?
A model where software cost depends on the amount of capacity licensed, usually per terabyte of raw, usable or logical capacity.
What is node-based storage licensing?
A model where software cost depends on the number of servers running the software, regardless of the capacity each holds.
Which licensing model is cheaper at scale?
For fast-growing environments using dense hardware, node-based licensing often costs less per terabyte over time. Smaller environments may favor capacity licensing.
Why does drive size matter for licensing?
Under node licensing, larger drives add capacity without adding licenses. Under capacity licensing, more capacity always means more license cost.
What should I check in a storage license contract?
How capacity is measured, support pricing, escalators, feature fees, refresh terms, minimums and exit costs.