What is the combined workload?
Request packet rate, flow churn, subscriber count, shaping, mapping logs and failover reserve in the same test profile.
The strongest choice fits your subscriber model, hardware, operating team and migration path. Compare the whole deployment.
Showing all 7 evaluation areas.
| Evaluate | LiquidNAT approach | Dedicated appliance approach | Assembled software stack |
|---|---|---|---|
| Infrastructure | Software-first; review the supported server and NIC profile. | Hardware and software procured as an integrated system; check upgrade options. | Components selected individually; compatibility work needs ownership. |
| Operational context | Gateway, RADIUS and Manager organised around the subscriber journey. | Evaluate the management suite and service correlation. | Correlation depends on the tools and integration your team builds. |
| Identity & policy | Integrated FreeRADIUS architecture or an external RADIUS integration discussion. | Check attributes, external servers and service-change behaviour. | Flexible component choice; attribute and policy mapping need ownership. |
| Total cost | Scope software, servers, support, logging and migration together. | Include hardware refresh, capacity licences and support renewals. | Include engineering time, integration, maintenance and escalation. |
| Migration | Topology-led review, mapping, representative validation and staged transition. | Assess vendor tooling and exact source-platform coverage. | Custom import and validation workflows need maintenance. |
| Recovery | Discuss forwarding, session, configuration and site recovery separately. | Verify each guarantee against the topology and enabled features. | Component resilience requires end-to-end recovery validation. |
| Evidence | Request the applicable capability, hardware and test profile. | Request the same profile alongside product-family claims. | Reproduce tests across the exact assembled versions. |
These are architectural comparison questions, not universal claims about every vendor or deployment. Product versions, licences and supported configurations matter.
Compare scope, operations, integration and the evidence to request.
Request packet rate, flow churn, subscriber count, shaping, mapping logs and failover reserve in the same test profile.
Ask separately about the UI, control plane, RADIUS, database, node failure and site loss.
Ask for explicit mappings, unsupported settings, credential handling, rollback and accounting reconciliation.