SIP Trunk Renewal: Reconcile Every DID, Channel, and Failover Route

A SIP trunk can keep working long after its commercial and technical design stops matching the business.

Phone numbers stay attached to closed offices. Channel commitments reflect an old call pattern. Failover still points to a destination nobody tests. Emergency settings survive a relocation. A secondary trunk gets billed every month, but nobody knows whether production traffic can reach it.

Then the renewal arrives and the provider asks for another term based on last year’s quantity.

Do not renew from the invoice. Reconcile every direct inward dialing number, trunk, call path, capacity commitment, and recovery route against the voice environment you operate now. The decision is not simply renew or replace. You may need to resize, reroute, port, consolidate, correct, or investigate parts of the service before you sign anything.

Build a number and trunk register first

Start with the numbers. A provider summary that says you have 600 DIDs is not an inventory.

Export every number from the carrier portal, PBX, session border controller, UCaaS platform, contact center, fax service, alarm records, messaging platform, and billing system. Include toll-free numbers, main published numbers, direct lines, temporary numbers, and numbers retained for forwarding.

Use one row per number and record:

Field	What to capture
Number	Full DID or toll-free number
Business use	Main line, user, queue, fax, alarm, campaign, testing, or unknown
Location	Site and service address tied to the number
Routing	Current trunk, PBX, application, queue, or forwarding destination
Owners	Business owner and technical owner
Dependencies	Voice, SMS, fax, caller ID, emergency calling, integrations, or published listings
Activity	Recent inbound and outbound use, where available
Recovery	Failover route, last test, and test result
Commercial scope	Provider, account, rate center, recurring charge, and contract term
Decision	Keep, correct, port, retire, consolidate, or investigate

Match by number, not by count. Two exports can both show 600 DIDs while hiding 40 unused numbers and 40 production numbers billed under another account.

Unknown does not mean unused. Call the number, check recent call detail records, search public websites and internal directories, and ask the likely business owner. A number may receive only a few calls each month and still be the published contact for a legal notice, building entry system, medical office, or long-standing customer relationship.

If you plan to move numbers, treat porting as its own project. The CCaaS porting and rollback guide explains why a requested date is not the same as a carrier-confirmed cutover and why every production number needs an owner, route, test, and fallback.

Size capacity from real call records

SIP capacity is often sold through concurrent call paths, sessions, channels, or a provider-specific usage model.

Pull at least 90 days of call detail records. Use a longer period if your business has seasonal peaks, open enrollment, fundraising events, weather response, billing cycles, or marketing campaigns that materially change call volume.

For each trunk and major call path, calculate:

Peak simultaneous inbound calls
Peak simultaneous outbound calls
Combined peak, if the provider shares capacity across directions
Busy-hour patterns by day and location
Blocked, rejected, abandoned, or failed attempts
Average and high-percentile call duration
Contact center and automated dialing peaks
Conference, recording, and transfer behavior that may consume additional sessions
Expected growth, site changes, acquisitions, and planned application traffic

Do not size a renewal from average minutes. A trunk can look quiet across a month and still hit its limit every Monday morning.

Twilio’s Elastic SIP Trunking documentation gives a useful vendor example. It tells customers to estimate peak bandwidth from maximum simultaneous calls and notes that a trunk includes separate origination, termination, and number settings. That does not make Twilio’s design universal. It proves why a buyer needs call concurrency, network capacity, and routing configuration in the same renewal review.

Ask the provider how it counts a session during transfers, forwarded calls, call recording, conferencing, and failover. Ask what happens when you exceed the contracted level. The answer may involve call rejection, burst capacity, overage charges, throttling, or a different limit. Get the answer in writing and test the behavior in a controlled window.

The right number is not the lowest number you can defend on a spreadsheet. It is the capacity that handles expected peaks with an agreed safety margin and a known response when demand goes beyond it.

Trace the complete call path

A SIP trunk does not operate by itself. Calls may cross a carrier, public or private network connection, firewall, session border controller, PBX, contact center, recording service, identity system, and several routing tables before a person answers.

Draw the path for inbound and outbound calling. Then mark which party controls each hop.

For every trunk, document:

Provider trunk identifier and account
Origination destinations for inbound calls
Termination domain or addresses for outbound calls
Session border controllers and PBX clusters
Signaling and media ports, protocols, encryption, and allowed addresses
Caller ID and number formatting rules
Inbound number routing and outbound route selection
Codec and transcoding requirements
Network path, bandwidth, quality controls, and monitoring
Logging and packet-capture access
Support handoff between carrier, network provider, managed service, and PBX owner

Configuration drift creates ugly failures. A firewall rule may permit the primary provider edge but block another edge used during an incident. A certificate can expire. A network team can change routing without knowing the voice path depends on it. A managed provider can own the SBC while your internal team assumes the carrier owns it.

The broader UCaaS buyer’s guide covers platform selection, migration, support, and total cost. This renewal audit is narrower. It proves the trunk and its routes still match the phone system already in production.

Test failover instead of reviewing a diagram

A secondary route on a diagram is not evidence that calls will recover.

Provider behavior depends on the configuration. Twilio, for example, allows multiple origination SIP URIs with priority and weight values. Its documentation says lower-numbered priority values are tried first and other URIs can be used when a session fails. It also documents conditions where certain SIP responses will not trigger failover. That detail matters. A backup destination can exist and still never receive a call under the failure you expected it to cover.

Build a controlled test for each material failure mode:

Primary PBX or SBC becomes unreachable.
Primary site loses internet connectivity.
One provider edge or route is unavailable.
The business must forward calls to an alternate site, queue, mobile group, or recorded message.
Outbound calling must continue from the recovery environment.
The primary route returns and traffic must fail back cleanly.

For each test, record the expected trigger, who initiates it, which numbers are included, where calls should land, how long recovery may take, which logs prove the route changed, and how you will restore normal service.

Test representative main numbers, direct numbers, toll-free numbers, queues, transfers, voicemail, outbound caller ID, and any regulated recording path. Include a network and security reviewer so the test does not accidentally create an unapproved route around normal controls.

NIST SP 800-34 Rev. 1 says organizations should evaluate systems and operations to determine contingency planning requirements and priorities. Apply that principle to voice. Your executive line, patient scheduling number, service desk, customer support queue, and a rarely used conference room DID do not need identical recovery targets. Rank them by business impact and test accordingly.

Review emergency calling and fraud controls separately

Do not let emergency calling disappear inside a general routing test.

Match every site, dispatchable location, calling platform, number, and emergency service setting to the current environment. Confirm who updates location records after a move, how remote or nomadic users are handled, what notifications are expected, and what test method the provider supports. Use the emergency calling renewal audit for the deeper location and evidence review.

Fraud controls also deserve their own evidence. Review international dialing, premium destinations, concurrent call limits, authentication, source restrictions, unusual call alerts, after-hours rules, and escalation contacts. Ask who can block traffic, how quickly they can act, and what happens to legitimate calls during containment.

A renewal is a good time to remove rules created for a one-time project and to challenge wide-open dialing permissions nobody can explain.

Reconcile the technical truth with the contract

Once the inventory, usage, routing, and tests are complete, bring in the agreement, order forms, invoices, support records, and renewal proposal.

Check:

Every billed trunk, DID, channel, feature, route, and service location
Minimum commitments and burst or overage treatment
Usage rates and rating increments
Toll-free, international, emergency service, directory, and regulatory charges
Support boundaries and response targets
Fraud responsibility and dispute windows
Service-level commitments, exclusions, and credit process
Number ownership, porting authorization, and transition help
Configuration, call record, and invoice export rights
Notice periods, auto-renewal language, price changes, and termination charges
Dependencies on access circuits, managed routers, SBCs, or licenses under separate terms

Look for duplicated scope. A company using a cloud calling plan may still be paying an old SIP trunk because one fax machine, contact center flow, or forgotten PBX route was never migrated. Do not cancel it from a billing hunch. Trace the traffic, prove the dependency, and then remove or replace it deliberately.

If the renewal is already disputed, use the two-track telecom renewal plan. Resolve the billing evidence while protecting continuity and evaluating replacement options. One track should not wait for the other.

Make the decision trunk by trunk

Renew as proposed only when the numbers, capacity, call paths, failover, emergency settings, fraud controls, support model, and commercial scope all hold up.

Resize when the current commitment does not match measured peaks and expected change.

Correct or reroute when the service fits but the configuration, ownership, recovery path, or support handoff is wrong.

Port, consolidate, or retire when you can prove a number or trunk no longer belongs in the current design and you have protected every dependent workflow.

Compare alternatives when the provider cannot produce usable records, support your required recovery design, explain capacity behavior, protect the numbers, or offer terms that match how the business operates.

The renewal proposal is a price for the old design. Your job is to decide what the next design should be.

If your SIP trunk or voice agreement is approaching renewal, request a Contract and Spend Risk Review. Bring the agreement, invoices, number inventory, call detail records, concurrency data, routing configuration, failover test results, support history, and notice dates. Catch Advisors will help you decide what to renew, resize, reroute, port, consolidate, or compare before the current design rolls into another term.