IBM Power Sizing Evidence Checklist for IBM i and AIX
A deeper sizing checklist for experienced IBM i and AIX buyers who need to turn performance evidence into a defensible IBM Power hardware quote.
Move From Model Shopping To Evidence
A first-level IBM Power server guide can explain the model family. The next level is building the evidence package that lets a reseller, engineer, or internal reviewer size the purchase without guessing.
For IBM i and AIX buyers, the evidence package should show current workload behavior, not only the age of the current server.
- CPU peaks
- Memory pressure
- Disk busy
- Batch windows
- Partition layout
- Backup load
- Growth expectations
Collect Partition-Level Data First
Sizing should start at the LPAR level. A physical server may host several workload types, and each partition needs its own current and peak profile before shared server capacity can be evaluated.
- Production IBM i
- Test IBM i
- AIX
- Linux
- VIOS
- HMC-managed services
- Reporting or integration side workloads
- Active processors
- Capped or uncapped settings
- Memory allocation
- Disk latency
- Storage paths
- Network adapters
- Jobs that define peak windows
- Known workload seasonality
Separate Peak Events From Daily Load
Experienced buyers know the average workload can hide the real sizing problem. Peak events often decide whether a server feels fast enough.
The quote request should state which peak events matter and whether the business expects the new system to shorten them. Without that context, the quote may match average demand but fail the moments users actually complain about.
- Month-end
- Year-end
- Inventory close
- Overnight billing
- EDI bursts
- Report refresh windows
- How often each peak happens
- How long each peak can run
Turn Growth Assumptions Into Quote Inputs
Growth assumptions should be written as quote inputs, not vague expectations. A quote that includes three years of expected growth may cost more than a like-for-like replacement, but it is easier to defend because the headroom is tied to named business changes.
- Additional LPARs
- More users
- More batch work
- Higher transaction volume
- Extra reporting demand
- New storage drawers
- Future high availability project
What To Send Before Asking For Final Pricing
Send the quote package before asking for final pricing. That package removes friction from the quote process and reduces the chance that a reseller sends a model that looks right but misses the installed environment.
- Current machine type and serial number
- IBM i and AIX release levels
- HMC and firmware levels
- Partition list
- Performance summary
- Storage layout
- Backup window
- Adapter list
- Tape and console dependencies
- Power and rack constraints
- Required support window
Frequently Asked Questions
What evidence matters most for IBM Power sizing?
The measured workload evidence matters more than the current model name by itself.
- Partition-level processor use
- Memory pressure
- Disk busy
- I/O latency
- Storage layout
- Network adapter needs
- Batch windows
- Growth assumptions
Should sizing start from the current server model?
Use the current model as context, not as the sizing method. The safer approach is to measure current workload behavior and map it to the target Power 11, Power 10, or refurbished hardware option.
Related Resources
Sources
Product catalog pages tied to this Sizing topic.
AS400 Hardware Maintenance Plan
A maintenance and support plan for AS/400, iSeries, and IBM Power hardware where parts access, response time, and lifecycle risk matter.
Hosted IBM i Power Server Environment
Managed IBM i, iSeries, or AS/400 hosting on Power infrastructure for shops that need server capacity, support, and operations handled outside their own server room.
IBM P Series and AIX Server
A sourcing path for buyers using pSeries, System p, or AIX language while evaluating modern IBM Power server hardware.