Interkey Intelligent DPI: every question an operator asks

Yes, with an important qualification. It does not decrypt traffic. It identifies encrypted and tunnelled flows using their observable characteristics: handshake structure, packet size and timing patterns, endpoint behaviour, rather than by reading payload. This is how modern classification works, and it is why accuracy should be measured against your own traffic rather than taken from a datasheet.

No. It is stated by Interkey and is not independently benchmarked here. DPI accuracy depends heavily on the traffic it is measured against, so the meaningful test is a proof of concept against a sample of your own live traffic. Interkey will run one; so should any vendor you are comparing against.

Yes. Out-of-band deployment gives you classification and reporting without placing the platform in the forwarding path, which is how most operators begin. Enforcement requires an inline deployment, and that is normally a second phase once the classification has been validated.

Classification is only useful when it feeds enforcement, so integration with the operator’s policy and charging architecture is part of deployment design rather than an afterthought. Bring your policy architecture and the enforcement points you care about into the scoping conversation: the integration plan, and what it requires from each side, is established there and written down.

No. The platform is deployed on the operator’s own infrastructure and can run entirely in-Kingdom. Traffic does not need to leave the operator’s environment for classification to work.

Classification produces metadata: which applications, which volumes, which policies fired. What is retained, for how long, and who can query it are design decisions made against the operator’s own data-governance rules, and they belong in the deployment design and the contract, not in a vendor default. Ask every DPI supplier this question in exactly this form.

Six stages, in a fixed order: written acceptance criteria, a representative traffic sample and tap points, an out-of-band classification baseline, accuracy measured against the operator’s own ground truth, a policy dry-run, and only then inline enforcement and acceptance. The full sequence is on the main DPI page, written so it can be lifted into an RFP; durations are agreed at scoping, against your calendar rather than a template.

By deployment and capacity, on a quote basis. There is no published price list. The sizing depends on throughput, where it sits in the network and which of the policy and reporting capabilities are in scope.