Procurement

Buying protocol software from a vendor in another region: what to ask

2026-10-29 · 7 min read

Protocol software is a long-lived dependency embedded in a product you will support for a decade. When the vendor is in another country and another time zone, the questions that matter shift from "what does it cost" to "what happens at 2 a.m. when a customer site is down and the escalation path is asleep".

Licence scope: the four sentences to read

• What is delivered — source, static library, or both — and whether you may modify it.
• Where it may be embedded: one product, one product line, or unlimited products. This is where budgets diverge most.
• Whether redistribution of the compiled result is royalty-free, and what happens if volume grows by an order of magnitude.
• What happens on termination: can you keep shipping firmware you already built?

Support: measure the path, not the promise

Ask for the actual escalation procedure with names and times, not the marketing tier table. Then test it once with a non-critical question and see how long a substantive answer takes. The answer to that test is worth more than any SLA document.

• Which time zone does first-line support work in, and is there anyone on your side of the world?
• Who answers when the answer requires reading the source?
• Is there a channel for a production-down incident that bypasses normal ticketing?
• In which language will the answer arrive, and by whom written?

The engineering questions that reveal the vendor

• Show us the ports that exist and the ones that are "supported" but untested.
• What does your stack do when the peer sends a service you do not implement?
• Which optional features do you not implement, and how is that documented?
• Can we have a trial build and a test against our own SCL file?
• Who wrote the conformance test tool you use, and can we see its reports?

The contract terms people forget

• Escrow of the source, with a defined trigger and a defined release procedure.
• A stated maintenance period and what "maintenance" includes — new editions, bug fixes, or both.
• The right to fix and ship without waiting, if a customer site is down.
• A defined process for the vendor disappearing: who holds the source, and how do you get it.

Our own answers, for comparison

We deliver source under a per-project licence with royalty-free redistribution inside the buyer's product line, support in English and Chinese from Beijing (UTC+8), and a named engineer on escalation. Support for the protocol tooling is included because the tooling exists to answer exactly the questions above. If a vendor will not put their answers in the same form, that is informative.

Questions that come up

Is a cheaper licence with no support a false economy?

It depends on whether you have someone who can debug the stack. If you do, the licence price is the whole cost. If you do not, the first field escalation pays for the difference several times over.

How do we evaluate a vendor we cannot visit?

Ask for a recorded walkthrough of a defect being fixed in their stack, and a trial build against your own model. Both are hard to fake and cheap for a serious vendor to provide.

What if we need a feature the vendor does not have?

Ask whether it is on a roadmap with a date, and what it would cost to have it developed. "Roadmap" without a date is a polite no, and you should plan accordingly rather than build on the assumption.

Keep reading