If you download IPinfo databases in MMDB format, you may have noticed your files shrinking. Your data is fully intact. The files are smaller because of an optimization we rolled out across all our datasets.

The data inside is byte-for-byte identical, with every network, field, and lookup result preserved. Only the file size has changed, dropping by up to 81% for some datasets, and 32% across our entire catalog.

What Changed

MMDB is the de facto standard for shipping IP data. Its memory-mapped binary search tree answers lookups in microseconds with no external dependencies. It's also famously bulky. File size has always been the trade-off you accept for its speed and simplicity.

Your current integration reads the new files exactly as it did before, with the same format and the same tooling. The only change is a lighter file.

The Numbers

Across the 80 databases we generate daily, the optimization removed 24.4 GB, taking the catalog from 76.9 GB to 52.5 GB, a 32% overall reduction.

Not every dataset shrinks. The gain depends on how repetitive the data is. Datasets where many ranges share the same answer (location, privacy, tags) shrink the most; datasets where every range carries unique values (ASN, organization, WHOIS) were already efficient and stay the same size.

Database	Before	After	Reduction
ipinfo_lite	32 MB	22 MB	31%
ipinfo_core	1.3 GB	669 MB	51%
ipinfo_plus	5.0 GB	3.3 GB	33%
ipinfo_location	664 MB	304 MB	54%
location_extended_v2	1.1 GB	514 MB	54%
ipinfo_privacy	644 MB	121 MB	81%
ipinfo_privacy_extended	741 MB	251 MB	66%
resproxy_7d	3.3 GB	2.4 GB	27%
Total catalog (80 databases, incl. custom builds)	76.9 GB	52.5 GB	32%

What This Means for You

Faster downloads, lower ingress. Daily refreshes move a fraction of the bytes, which matters if you pull multi-GB files like Residential Proxy or Plus every day.
Smaller memory footprint, lower memory costs. MMDB files are memory-mapped, so smaller files mean less RAM on every lookup host. With memory prices having skyrocketed recently, those saved gigabytes translate directly into infrastructure savings.
Room to grow. Some platforms cap file sizes (a common one is 2 GB). Datasets that were pressing against those limits now fit comfortably, and we have headroom to keep adding IP context without making you re-architect.
Same or better performance. In our benchmarks, lookup speed is unchanged or slightly improved, since smaller trees mean fewer page faults.

Verified at Every Step

The optimization is lossless and requires no decompression at lookup time. Database contents and lookup results remain unchanged. Before rolling it out, every optimized database passed:

Full prefix enumeration, with every IP in the original matched against the optimized file in both directions
Validation with standard MMDB verification tooling, including 12 different MMDB SDKs
Benchmark parity on lookup latency and memory

The rollout has been live for weeks. Today, 100% of our API traffic and 99.9% of database downloads already serve the optimized files, with the same filenames and delivery you use today.

What's Next

We’ll share the engineering behind the optimization in a forthcoming technical deep dive. That’ll include a closer look at how we reduced MMDB file size while preserving compatibility and lookup performance.

We think the engineering community will enjoy it!

Questions about your specific datasets? Reach out to your account team or support@ipinfo.io.