How the hardware sizing calculator works
The calculator is currently provided in beta. Its estimates are based on internal testing and will be refined as more measurements and customer feedback become available, so results may change between versions.
Overview
The Hardware sizing calculator estimates the CPU, RAM and storage needed for a CAST Imaging deployment, including Imaging Services, analysis nodes, PostgreSQL, Neo4j, Imaging Viewer and shared folder storage. Use these estimates as an initial baseline when you provision your virtual or physical machines.
How the estimates are calculated
Estimates are based on:
- the number of applications
- the total lines of code
- the technologies used
- the number of large applications (L, XL, XXL and XXXL)
- how often you rescan
- the number of concurrent users
- optional application characteristics
Resource consumption also depends on factors that cannot always be predicted:
- the combination of technologies and analyzers involved
- application complexity, regardless of code size
- the analysis configuration and the options enabled
- resource-intensive operations, such as dataflow analysis, transaction evaluation and call-chain computation
- parallel analyses and the overall workload
Depending on your applications and usage, your environment may need more resources than the estimate.
Memory sizing on analysis nodes
The estimated RAM is a provisioning baseline, not a maximum memory limit. Memory consumption is harder to predict on analysis nodes than on other components, because it depends on the technologies analyzed, the application complexity and the operations performed.
Some analyzers have no fixed memory limit: their memory consumption follows the size and complexity of the application being analyzed and can grow up to the limits of the machine or container. Actual memory consumption can therefore exceed the estimate, particularly for complex applications. Treat the estimated RAM as a starting point, not as a guarantee that every analysis will complete successfully.
Container memory limits do not reduce the memory an analysis needs. If you set them too low, analyses can fail.
Running analysis nodes on separate machines from the database services prevents memory-intensive analyses from affecting other components.
Storage
Shared folder estimates cover CAST Imaging data only and exclude storage for your own source code. Plan additional capacity if source code or archives are stored on the same infrastructure.
Storage requirements also grow over time as you add, update and rescan applications.