Israel SIP Phone Planning

July 27, 2026 · View on GitHub

Planning workspace for moving personal, business and household telephony in Israel onto SIP, while keeping the existing phone numbers.

The report

A dated, self-contained PDF combining the specification, the evidence, the options and the implementation path — built with Typst from report/report.typ.

→ Download the latest PDF · rebuild locally with ./report/build.sh

Every factual claim in it carries a confidence marker — confirmed (read at the primary source), likely (consistent across independent secondary sources), unverified (single source, vendor marketing, or inference) — with sources and retrieval dates in Appendix B. The docs/ tree below is the live, continuously updated version; the report is a snapshot of it.

Contents

DocumentWhat it is
docs/spec.mdThe spec: objective, background, numbered requirements, open questions, scope
docs/decisions.mdDesign decisions and the reasoning that settled them
docs/research/2026-07-27-feasibility.mdFeasibility assessment with sources and confidence levels, and the recommended architecture
docs/research/2026-07-27-roaming-and-clients.mdRoaming for US trips, SIP client choice, and the two-subscriber case for a PBX
docs/implementation-plan.mdPhased plan — blockers first, irreversible steps last
docs/vendor-questions.mdCall script for Israeli providers, ordered so a bad fit ends the call in three minutes
docs/research/2026-07-27-portability-regulation-and-israeli-vendors.mdThe portability rule read from the regulation itself, and the bundled product it turned up
docs/research/2026-07-27-hosted-pbx-options.mdHosted PBX candidates and the shortlist
docs/costs.mdRate inputs and a modelled monthly cost at three call volumes
docs/provider-evaluation.mdGating and scoring criteria, and candidates assessed so far
docs/source/The dictated brief the spec was built from — cleaned and verbatim

Where this stands

The original objective — drop cellular entirely, run everything over SIP — is not achievable in Israel. There is no number portability between cellular and fixed/VoIP services, so an 05x number cannot move to a SIP provider; and 2FA, consumer WhatsApp and emergency calling each independently need the personal number to stay on a real mobile line. Twilio in particular cannot dial Israeli emergency services at all.

That is now confirmed from the Ministry of Communications regulation itself (§2(a) of the numbering plan for portability), not from Wikipedia. The same document also establishes three things that help: a port cannot be refused over a contract or an unpaid balance, a mobile port completes in under half an hour, and the number must stay live during the port because the receiving provider verifies possession by SMS or a call to it.

The architecture is one conventional cellular plan carrying the personal number, the data and international roaming (D1), with all SIP infrastructure built on top of it. Business and household lines are virtual. That models at roughly $26–57/month against $20–50 today.

The open decision: a single cellular line can hold only one 05x, and an 05x cannot move to VoIP — so what happens to the business number? Either it becomes a new 07x, or a bundled product holds both numbers on one subscription. Recorded in D1, not yet settled.

That bundled product turns out to exist. Voicenter — a mainstream Israeli vendor — sells a cloud PBX whose extensions can be an app, an IP phone, or a cellular SIM with data included, from ₪80/user/month. If that SIM carries a real, portable 05x, it is the only arrangement in which the personal number stays a genuine mobile line (so SMS, WhatsApp and emergency calling all still work) while its calls sit inside a PBX that can record them — which the DIY architecture cannot do. It would also plausibly take the business number, closing the D1 question.

None of that is verifiable from a web page, and all of it is disqualifying if wrong. Four questions settle it, and they are a phone call: docs/vendor-questions.md.

Travel is solved: Israeli SIM in slot 1 with data off for 2FA, US prepaid SIM in slot 2 for data. The slot D1 freed is what makes it work — no eSIM needed. SIP uses under 0.5 GB for a heavy three-week trip.

Hosted, not self-hosted (D3) — the home server could run it, but telephony availability should not depend on home internet and power. One subscriber for now (D4); the second person and the Home Assistant integration are deferred.

Next: work the plan from Phase 0, which now tests two architectures in parallel — Route A, the bundled Israeli vendor, and Route B, the DIY Twilio build that was previously assumed.

Start with one phone call to Voicenter. It is free, it takes twenty minutes, and it can collapse the entire fork: is the SIM extension a real 05x on a named mobile network, will they port the Pelephone number onto it, does it pass bank OTP and consumer WhatsApp, and will they sell one seat to a sole trader? Four answers. If any is "no", Route A closes and the plan reverts to B unchanged.

While waiting for the call back: test whether Zoiper push works against a Twilio SIP Domain from a foreign network (the deciding test for Route B), ask Pelephone what leaving costs, and confirm there is no porting block on the business number — the regulation lets a business account impose one, and it would fail the port silently at the moment it is attempted.