Most IBM i hosting providers do something that costs their clients thousands of dollars a year without anyone noticing. When a client moves to hosted infrastructure, the provider mirrors whatever the client brought in. That means setting the baseline on oversized hardware, excess memory, and storage nobody uses.

The right-sizing conversation never happens, and the client ends up paying for more than they need. For IBM i leaders evaluating hosting options, that gap between what the hardware can do and what your workload actually needs is where the real money is hiding.

Understanding the Gap Between IBM i Capacity and Usage

IBM builds Power servers based on manufacturing logic. It costs less per unit to produce more capacity. The smallest server a distribution company can buy today is a Power11. It ships with 29,000 Commercial Processing Workload (CPW) units, 64 gigabytes of memory, and 1.6 to 3.2 terabytes of storage as standard configurations.

Twenty years ago, those same workloads ran on servers delivering 400 to 4,000 CPW. The software running on today’s modern Power hardware was largely written during the 1980s and 1990s using RPG. IBM designed the programming language to be processing-efficient because compute capacity was expensive. That efficiency is still baked into the code, but the servers got bigger.

A company running 150 users on a Power9 rated at 13,000 CPW may only consume 2,000 to 3,000 CPW of actual working performance. Not because the server is slow. But because storage device count and drive speed create a ceiling well below the headline number. Think of it as a 64-lane freeway with five cars on it. You’ve built all those lanes, but it makes no difference to the flow of traffic.

Inside an IBM i LPAR Right-Sizing Assessment

The alignment process begins with our operational assessment form. The Source Data team requests hardware reports, software profiles, and user counts during planning. The goal is to align your environment with precise operational workloads rather than just matching your existing hardware specs.

The analysis examines memory allocation alongside processing requirements and storage capacity. Organizations frequently run large memory configurations because manufacturers historically mandated minimum purchase requirements. Highly efficient software written years ago continues to run perfectly with smaller memory footprints today. Cutting memory to match your actual usage is smart engineering, not a compromise.

The CPW conversation is more nuanced and requires specialized and experienced operational insight. Familiarity with ISV platforms enables precise system-loading calculations. Custom code typically remains lean and highly efficient during daily business operations. ISV software carries more overhead by design because it is built for broad functionality rather than narrow efficiency. That distinction matters when sizing a logical partition.

The result is a recommended LPAR configuration smaller than the one the client brought in, while still meeting operational needs. That means it’s also less expensive to run than one that just mirrors existing hardware specs. Clients can also use a 60-day Cloud400 trial to demonstrate results under real-world conditions.

Proven Results From Real IBM i Clients

A busy home builder migrated their operations, seeking an upgraded and highly supportive hosting partnership. The Source Data team completed an analysis and matched their precise workload to an optimized partition. We delivered comparable performance at about $1,000 less per month than their previous provider.

A municipality client presented an entirely different challenge in their transition needs. No one at the organization had the necessary IBM i knowledge, and the consultant assisting with the migration was 4 hours away. With limited visibility into actual workload data, the environment was sized based on the software platform and scaled proportionally to the user count. The system came up performing as expected.

A Washington, D.C., organization running a vintage 9406-520 at V5R3 needed to vacate its building on a compressed timeline. The machine had 8 gigabytes of memory and 300 gigabytes of storage, with only 150 gigabytes in use. We sized a Power9 LPAR at the smallest available increment, which is 1,300 CPW.

Our team used a direct application push approach rather than IBM’s standard multi-step upgrade path. Total consulting cost for the cutover ran approximately $2,000 to $3,000. That’s compared to the $15,000 a traditional upgrade process would have required. We had everything live in 30 days.

The Source Data Right-Sizing Advantage

Most IBM i practitioners know one industry deeply, or one application family. Very few have worked across manufacturing, distribution, logistics, hospitality, pharmacies, and agriculture. Fewer still have done the work to understand the relationship between CPW ratings and actual storage throughput. That’s a gap that IBM’s own documentation does not make easy to close.

Bob Losey has spent more than four decades building that cross-industry application knowledge through direct client work. That is what makes the pattern he describes so consistent across the market.

“I don’t know of another hosting company that does this. More often, what I run into is the customer calls up and says, ‘I have a Power9. It has 13,000 CPW. I have 64 GB of memory. I have 3 terabytes of storage. Give me a price for hosting that.’ What I’ll do is say, ‘Look, I’m going to show you what you probably can live with, that’s going to be more than adequate, and show you the price of that compared to what you’re asking for, and I’ll prove it to you with a free trial.’ Consequently, if we can right-size them and put them into the right kind of logical partition, they’ll generally find that we’re anywhere from 30 to 70% less expensive than the other candidates.” – Bob Losey, Founder and President, Source Data Products

Organizations that price their hosting options based on historical hardware specifications often unintentionally double their required budgets. Properly identifying the financial gap creates massive, permanent savings with no expiration dates.

We touched on these issues in a post about right-sizing in the cloud vs. on-premises hardware.

Common Questions About IBM i LPAR Right-Sizing

What is CPW and why does it matter for IBM i hosting costs?

CPW stands for commercial processing workload. It is IBM’s benchmark for measuring how much business application activity a Power server can handle. It affects hosting costs because most clients pay for a CPW rating tied to the hardware they brought in, not the CPW they actually consume.

How much can right-sizing reduce IBM i hosting costs?

Clients who move from an oversized hardware spec to a properly matched LPAR typically see cost reductions of 30 to 70 percent. The total savings depend on a range of factors. Those include how oversized the original environment was, which ISV software is running, and how efficiently the underlying RPG code was written. A workload assessment before quoting is the only way to determine where a specific environment lands in that range.

Does reducing memory or CPW affect IBM i application performance?

Not when the reduction matches actual usage. Most IBM i environments run software written during the 1980s and 1990s, when RPG was designed to be processing-efficient because compute was expensive. That efficiency is still in the code. The goal is to size memory to match actual demand rather than the minimum configuration shipped by IBM. When you do that, it typically has no measurable effect on end-user performance.

Why do most IBM i hosting providers not right-size their clients?

Many providers accept existing hardware specs to avoid conducting complex technical analysis. That avoids a more complex conversation and reduces the provider’s risk of making a mistake. The downside for the client is that oversized hardware becomes oversized hosting. That means the excess cost continues indefinitely, with no natural point of correction.

What is a Cloud LPAR and how does it differ from a dedicated hosted server?

A Cloud LPAR is a logical partition on a shared Power server. It has a defined slice of CPU, memory, and storage allocated to one client’s workload. Unlike a dedicated server, an LPAR can be sized precisely to match actual demand rather than the smallest available physical box. That makes precise right-sizing possible in a hosted environment.

How does Source Data validate a right-sizing recommendation before a client commits?

Source Data offers a 60-day trial on Cloud400. The recommended LPAR configuration runs under real production conditions before the client makes a long-term commitment. If performance does not meet expectations, we can adjust the configuration. This structure removes the risk that typically makes clients reluctant to move away from familiar hardware specs.

Is right-sizing relevant if we are already in a hosted IBM i environment?

Yes. When organizations move to hosted IBM i, they often just replicate their on-premises hardware profile. If that original profile was oversized, the excess cost carries over. A workload assessment can identify whether the current hosted configuration matches real demand. Just because you’re in a hosted environment doesn’t mean you’re right-sized.

What information does Source Data need to complete a right-sizing assessment?

The assessment form requests hardware reports, software profiles, user counts, and location data. For clients running ISV platforms like JD Edwards, BPIX, or Infor-lineage products, the ISV’s known load characteristics inform the analysis alongside the reported hardware data. For fully custom RPG environments, user count and batch window data carry the most weight.

Set the Right Baseline

IBM i hosting costs stay high when providers accept existing hardware specs without question. Those specs rarely reflect real usage, and the difference has a financial impact. A proper assessment resets that baseline to something grounded in actual workload demand.

Source Data identifies that gap through a structured workload assessment and real application insight. Each recommendation reflects how your system behaves under real conditions. Reach out to the team to see what right-sizing would change in your environment.

Bob Losey is the founder and president of Source Data Products. He has worked with IBM midrange systems since 1979 and served as a Premier IBM Business Partner continuously since 1992, guiding more than 4,000 organizations through hosting, migration, and disaster recovery decisions for IBM i environments.