NEXIAWhy operators look for a replacement
VOS3000, by Linknat, has been the default wholesale transit softswitch across much of Latin America, Asia and Africa for over a decade. It carried — and still carries — an enormous amount of traffic. The reasons an operator starts looking elsewhere are almost always the same four, and only one of them is technical:
- The supply chain. A large ecosystem of modified installers, "free" RPMs and licence generators circulates around the product. Anyone running one of those copies does not know what else came inside the bundle.
- Reseller lock-in. Many installations were not bought from the vendor but from an intermediary who is, in practice, the only person who knows how that server is put together.
- Pace of change. The design belongs to its era. What is expected of a switch today — an API for everything, a customer portal, margin visible on every CDR — arrives late or arrives patched on.
- The accounting black box. When a customer's balance can be edited in the database without leaving a trace, you do not win the argument with that customer using data.
The hard, public data point: in 2020 ESET researchers documented CDRThief, Linux malware written specifically to target Linknat VOS2009 and VOS3000 softswitches and exfiltrate call metadata — CDR records with times, duration, source and destination. This is not forum hearsay: it is published research, and it describes a platform family widespread enough that someone wrote bespoke malware for it.
None of this makes VOS3000 a bad product: the vendor's software is legitimate commercial software. What is true is that the risk concentrates in how it reached your server, and that question — who compiled what you are running — does not always have an answer.
What is actually worth comparing
The useful comparison is not feature lists: both will have rates, gateways and CDRs. It is the five things you feel while operating:
| What to ask | NEXIA Switch | Common in the market |
|---|---|---|
| Provenance of the running binary | Ed25519-signed bundles, verified against a pinned key before install | RPM of varying origin depending on who sold it |
| Price | Published on the site, no negotiation | Depends on the reseller and the country |
| Trying it without buying | Free and permanent up to 100 channels, complete platform | Scheduled demo or a borrowed copy |
| Balance integrity | Append-only ledger; the database rejects UPDATE and DELETE on entries | Balance in an editable column |
| Public surface | Exactly one: the admission proxy. Media server, database and panel on loopback | Varies with how whoever installed it left it |
| Per-call margin | Cost, sell and margin on the same CDR row | Computed by exporting and cross-referencing |
What you carry over from the old install
The move is scarier than it is expensive, because there are four things to bring and three of them are spreadsheets:
- Buy and sell rates. They export as CSV or Excel from the old switch and import by prefix. This is the bulk of the work and it is mechanical.
- Customers and their connection points. Authorised IP or SIP user, channel cap and opening balance. They get recreated; there is no standard format that migrates itself.
- Carriers and routes. The dial peers get rebuilt with the new switch's prefix logic, which is rarely identical.
- CDR history. Archive it, do not migrate it. Keep the export: the new switch has no business understanding the old schema.
How to do it without risk: do not cut over at once. Stand the new switch up in parallel, send it one test customer and one carrier, compare the billing on that traffic against what the old one would have charged, and only then move the rest in batches. The free 100-channel plan exists precisely so that stage costs nothing.
What will not change, and is worth knowing
To be fair to the comparison, there are things a new switch does not fix:
- Your carriers are still your carriers. If the route quality is bad, no LCR improves it — it only lets you pick another one faster.
- Fraud does not vanish, it gets contained. Blocking risky destinations by default prevents the classic surprise invoice, not everyone else's creativity.
- Migrating costs your time. A weekend of spreadsheets, with the old switch still up, is a realistic estimate for a mid-sized operation.
What NEXIA Switch offers
NEXIA Switch is a self-hosted Class 4 softswitch, written from scratch for the job VOS3000 does today in many operations: hardened SIP admission at the edge with Kamailio, a routing and rating core compiled in Rust, LCR with a minimum margin, real-time billing on an immutable ledger and anti-fraud active from install.
It installs with one command on a clean Ubuntu 24.04, the installer verifies the bundle signature before touching the server, and the free plan of 100 concurrent calls is permanent: enough to run the migration in parallel without rushing and without paying two licences in the same month.