The IP Geolocation Database for the Environments an API Can’t Serve
When every transaction hinges on knowing where an IP resolves — and your SOC, fraud, or risk pipeline can’t afford a network hop to find out — the geolocation lookup has to live next to your code. Not behind a request.
- Outbound HTTP dependencyYour enrichment uptime is somebody else’s uptime.
- Crosses your perimeterEvery lookup is an outbound packet leaving your boundary.
- Third-party SLAContract terms you didn’t write, on a control your auditors care about.
- Per-request latencyA network round-trip sits inside every hot-path decision.
- IPinfo
- 99.5%
- of locations fall inside RTT-derived physical bounds
- Competitors
- 20%
- of locations violate physics when measured the same way
Three Environments Where an API-Based Lookup Doesn’t Fit.
An API is the right tool for plenty of geolocation work. But some environments carry constraints an outbound lookup can’t satisfy.
Latency
Every API lookup is a round trip: your system asks, the network carries the question out and the answer back. At the speed and volume real-time systems run at, that round trip is time you don’t have. When the lookup sits in the critical path, the delay means a login goes unchecked, a fraud signal lands too late to act on, or traffic is routed before its location is even known.
External Dependency
An API lookup places a step in your critical path that runs outside your walls. Even when it works perfectly, part of your pipeline now depends on something you don’t operate. A local database removes that step: the lookup completes inside your environment, start to finish. During an active incident investigation, the geographic dimension of your analysis is never left waiting on a call leaving your network.
Data Residency and Compliance
Some environments can’t send an IP address out for a lookup at all. Financial services, government networks, and air-gapped infrastructure operate under rules or physical isolation that an outbound call simply can’t meet. Without a local option, these environments either go without geolocation context entirely or accept a compliance risk on every lookup.
What Geolocation Failures Cost Downstream
When geolocation lags, fails, or gets blocked, the cost extends beyond a single lookup. It’s measured in the decisions your systems make without it. And the gaps compound: a missed detection becomes a delayed response, becomes a compliance exposure, becomes an investigation you can’t reconstruct.
Missed signals
A login from Bucharest reads as San Francisco. The rule that should have fired didn’t.
Delayed decisions
Geo-restriction enforcement waits on a network hop. By the time the check returns, the session is already three steps in.
Compliance gaps
Data leaves a regulated region for a lookup that should have stayed local. Your DPO finds out from the audit.
Investigation blind spots
The IP that mattered to the incident no longer resolves to the same location through the API. Forensics can’t reconstruct what the system actually saw at request time.
Why the Usual Alternatives Fall Short
If you’ve already addressed geolocation another way, it’s worth asking what that choice is costing you in accuracy, effort, or exposure.
Free / open-source
Most free datasets are a stripped-down version of a paid product. They have deliberately limited accuracy, coverage gaps, and slow updates. IPinfo Lite is the exception: accurate at country level, free, and licensed for commercial use with attribution. The full geolocation database is what you move to when you need more than country level: city-level precision, the complete field set, and enterprise support.
API-dependent
Every constraint above still applies. You’ve changed the provider, not the model: the lookup still leaves your network, still adds latency, and still can’t run where the data isn’t allowed to go.
Legacy / homegrown
Building your own works until it doesn’t. Someone has to keep it current, accuracy drifts the moment they stop, and the engineering time to maintain it rarely shows up on a budget line until you total it up.
Want to see the data in action? We’ll show you exactly what the gap looks like on your traffic.
The Geolocation Database: Local Lookups Without External Dependencies
Offline IP geolocation runs entirely inside your environment, with no outbound call, no rate limits, and no provider to depend on at request time. You query a local file the way you’d query any dataset you already own, and the answer returns in under a millisecond. It works in air-gapped networks, behind strict egress controls, and at the line-rate volumes an API was never built to serve. Delivered as a local MMDB or CSV file, it drops into the systems you already run.
- Local lookups
- No HTTP
- Air-gapped friendly
- Same answer everywhere
See it on your own traffic
Pull a 10,000-row sample, run it against last week's logs, see where it agrees with what you have and where it doesn't.
What the data covers
Every record resolves an IP to a precise location, with the fields your systems need to act on it.
Accuracy you can verify
Every location we publish is physically possible: measured against real-world network signals from ProbeNet, not estimated. It’s geolocation built on evidence you can inspect, not a percentage you take on faith, and no other provider works this way. Test it on the IPs that matter to you, or talk to sales about a proof of concept on your own data.
Probe Network
Our internet measurement platform, hosted on real ISPs in every populated region we serve, across the globe. Each probe reports a verified ground-truth location.
Ground-truth verification
Every IP block in the database is checked against probe-confirmed coordinates weekly. Disagreement above threshold opens an investigation before the next snapshot ships.
Continuous refresh
Daily full snapshots and hourly deltas. When accuracy drift is detected on a prefix, an out-of-band correction goes into the next delta — not the next monthly release.
Switching Geolocation Databases Without Rebuilding Your Stack
One catalog. Four formats. Daily refresh. Pick the one your stack already speaks. If you’re already running a local geolocation database, switching to IPinfo is a file swap: same place in your stack, better data underneath.
MMDB
~600 MBMemory-mapped binary. Sub-millisecond reads, used in production by an independent benchmark.
CSV
~1.8 GBRange-per-row. Drop straight into a warehouse or grep from a script.
JSON
~2.4 GBOne record per IP block. For pipelines that already speak JSON.
Parquet
~480 MBColumnar, compressed. The format Spark, DuckDB, and BigQuery want.
Cloud storage push
AWS S3, Google Cloud Storage, or Azure Blob Storage. Pulled into your account via presigned URL. IAM-scoped, versioned, region-pinned.
Direct download
Signed HTTPS URL with a rotating token. Drop into a cron, ship to your CDN, mirror to your registry.
Webhook
POST to your endpoint the moment a new snapshot is signed and ready. Useful for time-sensitive enrichment.
Cloud marketplace
Subscribe through Snowflake Marketplace or Google Cloud Marketplace. Procure on existing cloud commit, data flows into your account.
The file lands in your environment. From there, it's a connector or a one-line load into whatever your team already runs.
Migrating from MaxMind?
We offer a MaxMind-compatible variant of the database, so moving over is a matter of mapping the schema, not rebuilding your pipeline.
IPinfo’s high-quality data and complete contextualized IP insights have significantly improved advertising revenue for our customers, and the seamless integration with IPinfo’s databases via API led to a smooth onboarding process for our development team.
We ended up evaluating two different vendors where IPInfo proved to be better both in terms of overall performance (number of correctly identified VPNs) and data labeling.
IPinfo’s high-quality data and complete contextualized IP insights have significantly improved advertising revenue for our customers, and the seamless integration with IPinfo’s databases via API led to a smooth onboarding process for our development team.
We ended up evaluating two different vendors where IPInfo proved to be better both in terms of overall performance (number of correctly identified VPNs) and data labeling.
Not Sure You Need the Database?
If your volumes are modest or your lookups are occasional, the API or a standard subscription may serve you better: the same IP geolocation data, delivered differently. And if country-level accuracy covers your use case, IPinfo Lite is free and licensed for commercial use with attribution; the full geolocation database is for when you need city-level precision, the complete field set, and enterprise support. We’d rather route you to the right fit than sell you more than you need.
See the Data Before You Commit
Test it against your own IPs, in your own environment. When you’re ready to license, our team will scope coverage, formats, and delivery with you.
Frequently Asked Questions
For most stacks the switch is a file swap. IPinfo ships in the same MMDB format the GeoIP2 ecosystem uses. The difference is the data underneath: ours is measured against real-world network signals through ProbeNet and refreshed daily, and a MaxMind-compatible schema (with geoname_id) is available so migrating means mapping fields, not rebuilding.
We don’t answer that with a single percentage, and it’s worth being skeptical of anyone who does: industry accuracy numbers rarely share a definition or a method, so they’re hard to compare and easy to game. Our approach is to validate every location against real-world network measurements so the result is physically possible, then give you the tools to check it yourself. Download a sample and test it against IPs you already know, or talk to sales about a proof of concept on your own data. Accuracy you can verify is the only kind that matters for a database you’ll run in production.
The database refreshes daily, and you choose the cadence you ingest it (daily, weekly, or monthly) based on how sensitive your use case is to change.
Yes. Lookups run against a local file with no outbound call, so the database works in fully air-gapped networks and behind strict egress controls. The data lives wherever your systems can reach it on your own network.
MMDB for fast binary lookups, CSV for database imports, and Parquet and JSON for data-lake and warehouse pipelines. It’s the same data in whichever format your stack expects.
You query a local database file directly, the same way you’d query any dataset you host. There’s no network call, no rate limit, and no key to manage; the lookup happens entirely inside your environment in under a millisecond.
MMDB is a compact binary format built for fast IP lookups. It’s the same format the GeoIP2 ecosystem uses. You read it with a standard MMDB library in the language you already work in, pass an IP, and get the location record back.
Yes, with the right license. If you’re embedding geolocation into a product you ship to your own customers, talk to our team about OEM and redistribution terms. Standard subscriptions don’t include redistribution rights.
Yes. Parquet and JSON are built for warehouse and lakehouse pipelines, so you can join IP geolocation data with the rest of your datasets in Snowflake, Databricks, or BigQuery and enrich at query time.