Using a SIM Bank for Global SIM Management

A global SIM estate is not one estate. It is a set of national arrangements administered through one place, and the distinction decides the topology, the register and the reporting.

This article sets out what changes when the estate spans countries, how registration requirements become design inputs, where the estate and the radios should sit, and how reporting has to work when the people reading it are in different time zones.

What changes for a SIM bank when a global SIM management estate spans countries?

The subscription is local; the administration is central.

Nothing about the radio changes across a border, and everything about the commercial and regulatory relationship does.

Three things change. The number and the subscription belong to the operator that issued them, so a card issued in one market cannot present an address in another. The registration requirements differ, so a subscription that is sufficient in one market may not be in the next. And the operator relationships are separate, so a problem with one does not affect the others while a commercial change in one does not solve anything elsewhere. A deployment that plans around a single global pool will discover all three at the first expansion.

The identifiers that make a global register possible are standardised internationally, which is what allows one register to describe several national estates. The network identity is described in ITU-T E.212, the card identifier in ITU-T E.118, and the number in ITU-T E.164. Keying the register on all three, with the market recorded as a primary attribute, is what makes a global estate administrable rather than merely large.

Attribute Why it is a primary field What it decides
Market Subscriptions are national Which operator and terms apply
Card identifier Identifies the physical or profiled card Inventory and replacement
Network identity Identifies the issuing operator Roaming and routing questions
Number What recipients see Identity rules and reply routing
SIMBANK128 centralised SIM bank unit with 128 SIM slots
SIMBANK128, published at a list price of $1,600; central administration and local radios are independent decisions.

Registration requirements as a design input

They decide where the estate can sit, not only what it may send.

Requirements around subscriber identity, permitted traffic and equipment registration differ by market and come from the operator rather than from a specification.

Treating them as a design input means collecting them before the hardware is specified, because they can change the topology. A market that requires the subscriber to be identified may rule out an arrangement that works elsewhere; a market that restricts the traffic profile may make a subscription unsuitable for the workload the deployment intends. Collecting those answers in writing per market is the work that prevents a deployment from being built and then found unusable.

See also  SIM Card Management for SMS Modems: Rotation, Cooling & Safety

The registration states a subscription passes through are standardised, and they are defined in 3GPP TS 24.301. What differs by market is what has to be in place before the subscription is allowed to reach them, and that is the part worth confirming with the operator in each country before cards are ordered. The message path that follows is defined in ETSI TS 123 040.

Where the estate sits versus where radios sit

Centralise the cards, localise the radios.

The estate can be administered in one place while the transmission happens in each market, which keeps the network relationship local and the administration central.

That arrangement is what a SIM bank or SIM pool makes practical: cards held centrally, allocated to radios in the markets they serve, and reassigned administratively rather than physically. The benefit is administrative rather than technical, and it is largest where the estate changes often. The cost is a dependency on the link between the estate and the radios, which carries command traffic and therefore has a latency budget that the deployment has to respect.

Where that latency is unacceptable for an operation, the honest answer is to issue that operation locally rather than to accept a link that makes it unreliable. A global estate can hold both arrangements, provided the register records which sites are on which model. Deciding per site rather than per fleet is the same principle that applies to redundancy: the right answer depends on what the site does.

Block assignment by market or customer

Allocation is a policy, and it should be visible.

Cards can be assigned in blocks by market, by customer, or by service, and the choice determines how a problem propagates.

Blocks by market keep a commercial problem in one country, which is usually the property that matters most for a global estate: a dispute with one operator should not affect traffic in another. Blocks by customer make attribution easy and concentration risk high, because one customer’s block is a single administrative unit. Blocks by service are useful where the same market carries several products with different requirements.

The practical approach for most global deployments is a two-level assignment: a block per market, and within it an assignment per customer or service. That gives the isolation of the first and the attribution of the second, and it makes the register a small hierarchy rather than a flat list of thousands of cards. The allocation policy should be documented, because it is the kind of setting that gets changed casually and produces an attribution problem months later.

SK SIMPOOL 512 centralised SIM storage unit for a global estate
SK SIMPOOL 512, published at a list price of $5,400; the estate is administered centrally and remains national in its subscriptions.

The link between estate and radios across sites

It is the dependency that a global topology introduces.

Every operation that reaches across the link inherits its latency and its availability, and the ones that matter are the interactive ones.

Three operations cross the link in a typical deployment: allocation, command and reporting. Allocation tolerates latency because it is rare. Reporting tolerates latency because it is batch. Command is the one with a budget, because a command that is part of a live operation has to complete before the operation moves on. Where the link cannot meet that budget, the operation should be issued locally, which means the local site needs enough capability to act without the estate.

See also  How to manage thermal conditions in Southeast Asian4G SMS gateway setups?

The link also has a failure mode worth planning for. Where it is unavailable, the estate is present and the radios cannot use it, which is a different situation from the radios being down. The runbook should distinguish the two, because the first is an administration problem with a workaround and the second is an outage. Record-keeping practice for the logs that establish which occurred is covered in NIST SP 800-92.

Reporting that works across time zones

One clock, stated in the report.

A global estate produces events in many local times, and a report that does not state its reference clock is a report that will be misread.

The workable convention is to store every event with a single reference time and to record the local time and the market alongside it. That gives a global view that can be aggregated without arithmetic and a local view that matches what the operator experienced. Where the reporting tool converts between zones, the conversion should be explicit rather than implied, because a market that observes daylight saving and one that does not will otherwise produce a one-hour discrepancy twice a year that nobody can explain.

Two reports are worth producing per market: the delivery rate by destination area, and the per-number volume against the market’s policy ceiling. Together they show whether the estate is being used as intended in each market, which is the question a global review actually asks, and neither requires aggregating across markets to be useful.

A topology outline

The outline starts with the markets and works outward to the topology that serves them, rather than the other way round. Each step produces a decision about where the cards sit, how the radios reach them, and how the reporting handles several local times.

  1. List the markets, and for each one the operators that can issue a subscription.
  2. Record the registration and identity requirements per market before ordering.
  3. Decide where the estate sits and where the radios sit, per site rather than per fleet.
  4. Define the allocation hierarchy: a block per market, an assignment within it per customer or service.
  5. Measure the link latency for allocation, command and reporting separately, and issue locally where command cannot meet its budget.
  6. Store every event with one reference time and with the local time and market recorded.
  7. Produce delivery and per-number volume reports per market.
  8. Record the fallback behaviour when the link is unavailable but the estate is not.

The outline produces a global estate that behaves like a set of local ones with shared administration, which is what a global deployment actually is. Where a single-number global proposition is offered commercially, the arrangement behind it is a roaming or profile agreement rather than one subscription that works everywhere, and the same per-market documentation applies.

See also  How can optimal power and cooling be achieved for a32-port GSM gateway rackmount deployment?

Two further practices keep a global estate administrable. Record the operator relationship per market with its review date, because the commercial terms change and a plan that describes last year’s agreement is worse than no plan. And keep the register exportable in full, so that a market can be extracted and reviewed on its own without an engineering exercise. The second matters more than it sounds: a global register is usually maintained by a small team, and its value depends on being usable by someone who is not a specialist in the platform that holds it.

Where a market is served through a partner rather than directly, record the partner as well as the operator. The two relationships fail independently, and an escalation path that begins with the operator when the problem is with the partner adds a day to every incident.

Plan per market and administer centrally. Send your market list, allocation policy and site topology to service@telarvo.com, or review the published configurations on the SIMBANK128 page and the SK-SMS Gateway range. Telarvo publishes the SIMBANK and SIMPOOL ranges on its product pages, and the configurations referenced above come from those listings.

FAQ

Is there a SIM that works worldwide?

Cards and profiles roam across borders, and the number still belongs to the operator that issued it. A single-number global proposition depends on the operator arrangement behind it rather than on one universal subscription, and the commercial terms and acceptance behaviour differ by market. Confirm the arrangement in each market before designing around a single global proposition, because the terms differ and they change.

Should the whole global estate sit in one location?

The cards can be administered in one place while the radios stay in each market. That arrangement centralises administration and keeps the network relationship local. The dependency it introduces is the link between the estate and the radios, which matters for operations that must complete quickly. Measure the link under load, because the operational latency is what determines which tasks can be performed across it.

How should cards be assigned across markets?

In blocks per market, with an assignment within each block per customer or service. The block isolates a commercial problem to one country, and the assignment inside it preserves attribution. Document the policy, because changing it later changes what the historical reports mean. Review the block allocation when a market is added, because the original policy may not describe the new case.

What is the biggest reporting mistake in a global estate?

Not stating a reference clock. Events arrive in many local times, and a report that does not say which clock it uses will be misread, particularly across markets with different daylight saving rules. Store one reference time and record the local time and market alongside it. Record the reference clock in the reporting specification, because a report that omits it will be interpreted differently by each reader.

Your Guide to VOIP, SMS Gateways, and Telecom Trends - Telarvo Store Blog