Reverse DNS Database

The Reverse DNS Database contains the PTR record (the reverse DNS hostname) of every IPv4 address on the Internet that has one. It is the result of our own sweep of the complete routable IPv4 address space: 3,702,258,432 IP addresses queried, which returned 1,160,910,713 PTR records. You get the whole Internet's reverse DNS as a single download of 1.08 GB, ready for offline lookups, bulk analysis and research.

  • Every routable IPv4 address queried, not a sample
  • The DNS status of every address, including those without a PTR record
  • A forward-confirmed reverse DNS (FCrDNS) check of every hostname
  • IPv6 reverse DNS: 7,363,471 PTR records from a walk of the ip6.arpa tree
  • A compact binary format and a reader tool that converts it to TSV or JSON
  • A new crawl pass every one to two months

The database also covers IPv6. The IPv6 space cannot be enumerated, so our crawler walks the ip6.arpa reverse tree below every allocated IPv6 prefix and skips every subtree that does not exist: 7,363,471 individually named IPv6 PTR records in 69,027 /48 networks, plus 182 networks whose names are synthesized from the address, listed as patterns (see IPv6 coverage).

Reverse DNS hostnames are among the most informative signals that exist for an IP address. A single PTR record such as ec2-3-137-190-186.us-east-2.compute.amazonaws.com reveals the hosting provider, the service and the region of an IP address, while dynamic-077-183-045-122.77.183.pool.telefonica.de identifies a home Internet connection with a dynamically assigned address.

Reverse DNS Database

Price $149 per month
File size 1.08 GB (compressed)
Format .rdnsz, IPv6 in .rdns6z (reader tool included)
IPv4 addresses queried 3,702,258,432
PTR records 1,160,910,713
Updates Every one to two months (full sweeps and update passes)
Last crawl 2026-11-09
IPv6 PTR records 7,363,471 (+ 182 generated zones)
Last IPv6 walk 2026-09-05
Download Sample (.rdnsz) Sample (RDNSZ)
Download Sample (TSV) Sample (TSV)
Download Sample (IPv6) Sample (RDNS6Z) Sample (TSV)

# Use Cases

IP classification

Tell datacenter, residential, mobile, carrier-grade NAT and corporate addresses apart by their naming patterns, the same technique behind parts of the ipapi.is API.

Threat intelligence

Attribute scans and attacks to the infrastructure behind them. PTR records name hosting providers and their customers long before WHOIS records are updated.

Email deliverability

A missing, generic or unconfirmed PTR record is a classic spam signal. Check sending infrastructure at scale without a single live DNS query.

Attack surface discovery

Find every host of an organization by searching a billion hostnames for its domains, including forgotten servers in address ranges you did not know about.

Internet research

Study naming schemes, carrier-grade NAT, IPv4 utilization and infrastructure trends with a complete and consistent snapshot of the IPv4 reverse zone.

Offline lookups

The whole dataset is a local download of 1.08 GB: unlimited lookups, no latency, no DNS traffic, also in air-gapped networks.

Our article Enrich IP Data with Reverse DNS Lookups shows what reverse DNS data reveals in practice.

# Purchase Database

Create a free account, subscribe in your dashboard with a card or PayPal, and download the database right away, in every format it comes in. The subscription is billed monthly and can be canceled at any time. Updates are included: your personal download link always serves the latest release, so one scheduled command keeps your copy current. See pricing and the terms.

Need more than one database? The Business plan includes all seven databases ($563 per month if bought separately), the largest standard API plan and the right to run the API on your own servers, for $200 per month.

# Database Coverage

Unlike most reverse DNS datasets, this database does not only store the IP addresses that have a PTR record. It stores the DNS outcome of every queried IPv4 address, as one of these statuses:

  • has_ptr: the IP address has one or more PTR records. All hostnames are included.
  • noerror_empty: the authoritative nameserver answered, but returned no PTR record.
  • nxdomain: the reverse zone states that no PTR record exists for this IP address.
  • servfail: the responsible nameserver failed to answer the query.
  • refused: the responsible nameserver refused the query.
  • timeout: no answer was received, often because the reverse zone is not delegated at all.
  • net_error: a network error occurred while querying.
  • lame_delegation: the reverse zone is delegated to a nameserver that does not answer authoritatively.

The negative statuses are as valuable as the positive ones: nxdomain versus timeout tells a network operator that manages its reverse zones apart from address space without any DNS infrastructure, a strong signal for network hygiene scoring and abuse analysis.

For every PTR record, the database also stores the result of a forward-confirmed reverse DNS (FCrDNS) check: whether the hostname resolves back to the same IP address. Mail servers use the same check to judge a sender.

# IPv6 Coverage

IPv6 has too many addresses to query one by one, but its reverse tree has one label per hex digit, and an authoritative answer that a prefix does not exist means nothing exists below it (RFC 8020). The crawler walks that tree below all 518,390 allocated IPv6 prefixes of the regional registries and skips every empty subtree: the 2026-09-05 walk took 2,008,224,159 DNS queries. Since there is no list of addresses to begin with, the IPv6 part contains the addresses that have a PTR record: 7,363,471 of them in 674,689 /64 networks, without negative statuses.

Some operators synthesize PTR names from the address itself, for example dynamic-2a06-e184-d7ff-1--2.scarfbot.com. The walk collects such names until its per-prefix query budget runs out (255,612,483 of them), so they are an arbitrary sample of a network, not a list of hosts. The database lists these 182 /48 networks in ipv6/generated-zones.tsv instead, each with its naming pattern and an example. The largest:

Zone Name template
2801:88:fc1::/48usr-{h1}-{h2}-{h3}-{h4}-{h5}-{h6}-{h7}-{h8}.v6.v100.net.fclar.unesp.br
2a0f:5382:acab::/48dynamic-{h1}-{h2}-{h3}-{h4}-{h5}-{h6}-{h7}-{h8}.flpk.zentralwerk.org
2a01:110f:5f5f::/48{H1}-{H2}-{H3}-{H4}-{H5}-{H6}-{H7}-{H8}.ab.ipv6.supernova.orange.pl
2a01:110f:55d5::/48{H1}-{H2}-{H3}-{H4}-{H5}-{H6}-{H7}-{H8}.ab.ipv6.supernova.orange.pl
2a06:e184:d7ff::/48dynamic-{h1}-{h2}-{h3}-{h4}-{h5}-{h6}-{h7}-{h8}.scarfbot.com
2804:20c:4fcf::/48dynamic-{h1}-{h2}-{h3}-{h4}-{h5}-{h6}-{h7}-{h8}.projesom.net.br

{h1}–{h8} stand for the address's eight hextets in lowercase hex without leading zeros, {H1}–{H8} for the same zero-padded to four digits; operators that write the compressed form turn a run of zero hextets into --. A row says the names in that network are machine-generated, not that every address in it resolves.

# Database Format (.rdnsz)

Storing a billion hostnames as plain text takes more than 30 GB. This database ships in .rdnsz instead, a binary format built for it that stores the complete sweep, all 3,702,258,432 query results including every hostname, in 1.08 GB: less than a byte per queried IP address. It uses three techniques:

  • IP delta encoding: records are sorted by IP address, and only the difference to the previous IP address is stored, as a varint.
  • Hostname templates: the IP address embedded in hostnames like 40.64.0.1.megaegg.ne.jp is replaced by a placeholder, so millions of structurally identical ISP hostnames compress to almost nothing.
  • Zstandard block compression: records are grouped into blocks of 65,536 and compressed with zstd. Each block can be decompressed on its own, so the database can be streamed without loading it into memory.

The database is delivered as a .tar.gz package with 20 .rdnsz shard files (partitioned by IP range), the reader tool rdnsz_query.js, a machine-readable stats.json, the license and a README with the complete format specification, byte by byte, so you can write a reader in any language.

The IPv6 part sits in ipv6/: rdns-ipv6.rdns6z, a single file in the same block layout with a 16-byte address per record, sorted by address across the whole file, and generated-zones.tsv.

A converted excerpt of the database (IP address, tab, PTR hostnames):

1.0.16.108	st2-smtp.kakeibo.tepco.co.jp
1.0.64.20	20.64.0.1.megaegg.ne.jp
1.0.64.40	40.64.0.1.megaegg.ne.jp
1.0.64.60	60.64.0.1.megaegg.ne.jp
1.0.64.80	80.64.0.1.megaegg.ne.jp
1.0.64.100	100.64.0.1.megaegg.ne.jp
1.0.64.120	120.64.0.1.megaegg.ne.jp
1.0.64.140	140.64.0.1.megaegg.ne.jp
1.0.64.160	160.64.0.1.megaegg.ne.jp
1.0.64.180	180.64.0.1.megaegg.ne.jp
1.0.64.200	200.64.0.1.megaegg.ne.jp
1.0.64.220	220.64.0.1.megaegg.ne.jp
1.0.64.240	240.64.0.1.megaegg.ne.jp
1.0.65.4	4.65.0.1.megaegg.ne.jp

# How to Use the Reverse DNS Database

Every database package includes rdnsz_query.js, a self-contained reader tool without npm dependencies. It only requires Node.js 22.15 or newer, which has zstd support built in. To try it, download the sample database and the tool:

curl -O https://ipapi.is/data/samples/Reverse-DNS-Database-Sample.rdnsz
curl -O https://ipapi.is/src/rdnsz_query.js

Show the crawl statistics of a database file:

node rdnsz_query.js stats Reverse-DNS-Database-Sample.rdnsz

which outputs:

Reverse-DNS-Database-Sample.rdnsz: shard 0/1 | crawled 2026-11-09 | total=100,000 | has_ptr=10,863 noerror_empty=33 nxdomain=88,402 servfail=443 timeout=259 | sorted
---
files: 1
queried IPs: 100,000
  has_ptr                  10,863  (10.9%)
  noerror_empty                33  (0.0%)
  nxdomain                 88,402  (88.4%)
  servfail                    443  (0.4%)
  timeout                     259  (0.3%)
crawl date: 2026-11-09

Convert the database to plain text (TSV) or JSONL:

node rdnsz_query.js dump Reverse-DNS-Database-Sample.rdnsz --limit 5

which outputs:

1.0.16.108	st2-smtp.kakeibo.tepco.co.jp
1.0.64.20	20.64.0.1.megaegg.ne.jp
1.0.64.40	40.64.0.1.megaegg.ne.jp
1.0.64.60	60.64.0.1.megaegg.ne.jp
1.0.64.80	80.64.0.1.megaegg.ne.jp

Look up a single IP address:

node rdnsz_query.js lookup 1.0.16.108 Reverse-DNS-Database-Sample.rdnsz

which outputs:

{
  "ip": "1.0.16.108",
  "status": "has_ptr",
  "ptr": [
    "st2-smtp.kakeibo.tepco.co.jp"
  ],
  "fcrdns_match": false
}

The same tool reads the IPv6 file. It is sorted by address, so a lookup binary-searches it instead of scanning:

curl -O https://ipapi.is/data/samples/Reverse-DNS-Database-Sample-IPv6.rdns6z
node rdnsz_query.js lookup 2001:240:2432:f03::a Reverse-DNS-Database-Sample-IPv6.rdns6z

which outputs:

{
  "ip": "2001:240:2432:f03::a",
  "status": "has_ptr",
  "ptr": [
    "sri001kgigw2500.IIJ.Net"
  ],
  "fcrdns_match": false
}

After purchasing the database, you find your personal download command in the dashboard. The full database is downloaded the same way:

wget -O reverseDNSDatabase.tar.gz \
  'https://ipapi.is/app/getData?type=reverseDNS&format=rdnsz&apiKey=YOUR_API_KEY'

# What is Reverse DNS?

Normal (forward) DNS resolves a hostname such as ipapi.is to an IP address. Reverse DNS goes the other way: it resolves an IP address to a hostname by querying the PTR record of the IP address in the special in-addr.arpa zone. The PTR record of the IP address 1.2.3.4 is stored at 4.3.2.1.in-addr.arpa.

The network operator that holds an IP range controls its reverse zone. That is what makes reverse DNS so valuable: the hostname is set by the party that operates the IP address, and operators encode a lot into their naming schemes: provider names, service types (mail, vpn, nat, dhcp, static), locations, customer assignments and more.

The ipapi.is API derives several of its signals, such as is_datacenter, is_vpn and the detection of new hosting providers, in part from this reverse DNS data. With this database, you get the raw data to build your own detection logic.

# How the Data is Collected

The database is produced by our own distributed reverse DNS crawler:

  • Full coverage: every routable IPv4 address is queried, not a sample and not only "interesting" ranges. Private, reserved and other non-routable address space is excluded.
  • Distributed crawling: the sweep runs on a fleet of servers, each responsible for a fixed share of the address space.
  • First-hand answers: every crawl server runs its own recursive resolver, which asks the authoritative nameservers of each reverse zone directly instead of going through a public DNS resolver.
  • Retries: a query that fails gets three attempts of five seconds each before a negative status is recorded.
  • Verification: every hostname is resolved forward to check whether it points back to the original IP address (FCrDNS).
  • Targeted update passes: between full sweeps, the crawler re-queries only the addresses whose answer can change: those with a PTR record (to catch renames and removals) and those that failed with timeout or servfail (to recover records the previous pass missed). A known hostname is never erased by a single failed query; only an authoritative nxdomain removes it.

Because each release is a complete, self-contained snapshot, you can also compare two releases to detect infrastructure changes: new hosting ranges, ISPs renumbering their networks, or address space changing hands.

# Crawl History

Every crawl pass that went into the current database, newest first. IPs Queried is the number of PTR lookups performed in that pass (a full sweep queries the entire routable space; an update pass re-queries only the addresses whose answer can change). The status columns show the state of the published dataset after the pass was folded in, all relative to the 3,702,258,432 covered IPv4 addresses. An IPv6 PTR Records figure marks the release that first shipped an IPv6 walk.

Crawl Type IPs Queried PTR Records nxdomain timeout servfail Other IPv6 PTR Records
2026-11-09 – 2026-11-11update pass1.25B1,160,910,713 (31.36%)2.12B (57.15%)65.0M (1.76%)335.4M (9.06%)25.2M (0.68%)–
2026-09-05 – 2026-09-08update pass1.31B1,163,159,674 (31.42%)2.11B (56.87%)83.2M (2.25%)326.0M (8.81%)24.2M (0.65%)7,363,471
2026-07-29 – 2026-08-02update pass1.75B1,157,507,027 (31.26%)2.08B (56.24%)155.0M (4.19%)285.0M (7.70%)22.7M (0.61%)–
2026-07-02 – 2026-07-05full sweep3.70B1,039,377,899 (28.07%)1.94B (52.29%)561.4M (15.16%)146.9M (3.97%)18.8M (0.51%)–

The background of the crawls is documented on the blog: the July 2026 baseline sweep and the August 2026 update pass.

# Frequently Asked Questions

What is a reverse DNS database?

A list of IP addresses with their PTR records, the hostnames their operators assigned to them. Ours covers every routable IPv4 address: 1,160,910,713 PTR records from 3,702,258,432 queried addresses, with the DNS status of every address.

Why not run the reverse DNS lookups myself?

For a few addresses, a live lookup is fine. Querying all 3,702,258,432 addresses takes days on a fleet of servers, each with its own recursive resolver. The database gives you the result as a file, with unlimited offline lookups and without sending any DNS traffic.

How often is the database updated?

Every one to two months, with a full sweep or an update pass that re-queries every address whose answer can change. The crawl history above lists every pass in the current release.

Can I get the data as CSV or JSON?

Yes. The included reader tool converts the database to TSV or JSONL with a single command, and the format specification lets you write your own reader.

Is IPv6 included?

Yes: 7,363,471 IPv6 PTR records from a walk of the ip6.arpa reverse tree below every allocated IPv6 prefix, plus the networks whose names are generated from the address.

Do you offer custom exports?

Yes. Contact us, we are happy to help with integration or to provide a custom export.

All seven databases are also part of the Business plan, together with the API.