Derate a perimeter inspection stack stage by stage to find real capacity, added latency and unmonitored public endpoints.
Service chains compound multiplicatively, not additively: four stages each costing 15% leave 0.85⁴, or about 52% of rated capacity — not 40%. Latency, by contrast, is additive, so each stage adds its full delay to every session in both directions. The endpoints bypassing the stack matter more than the capacity number: one internet-facing service that routes around the perimeter makes the whole chain irrelevant for that path.
North-South Security
effective capacity = rated capacity × (1 − per-stage derate)^stages; added latency = stages × latency per stage.
effective capacity = rated capacity × (1 − per-stage derate)^stages; added latency = stages × latency per stage. Service chains compound multiplicatively, not additively: four stages each costing 15% leave 0.85⁴, or about 52% of rated capacity — not 40%. Latency, by contrast, is additive, so each stage adds its full delay to every session in both directions.
The endpoints bypassing the stack matter more than the capacity number: one internet-facing service that routes around the perimeter makes the whole chain irrelevant for that path.
This calculator takes 8 inputs: Peak ingress traffic, Peak egress traffic, Inspection stages in the chain, Throughput lost per stage, Rated perimeter capacity, Latency added per stage, Internet-facing endpoints, Endpoints behind the inspection stack. The pre-filled defaults are a realistic starting point — replace them with figures from your own environment for a result you can act on.
No. Each stage adds latency, a failure domain and an operational owner, and overlapping stages often detect the same things twice while nobody owns the gaps between them. Consolidating four single-function appliances into one platform that runs the same four engines usually gives better coverage at lower latency, provided you size the derate honestly.