DNS benchmark: how to find the fastest DNS server for your connection

A DNS benchmark measures how quickly different DNS servers look up domain names from your own internet connection. It tells you whether your ISP's resolver or a public DNS service such as Cloudflare, Google or Quad9 answers fastest and most reliably for you. You can run one with GRC's DNS Benchmark, with the free dnseval tool, or with a few commands in a terminal.

This guide covers what a DNS benchmark measures, which tools are available in November 2026, how to run a fair test and how to read the results. For the difference between resolvers and how to switch, see public DNS vs ISP DNS servers.

What is a DNS benchmark?

Every time you open a website, your device asks a DNS server (a resolver) which IP address belongs to the name. A DNS benchmark sends the same kind of queries to a set of resolvers and records, per server, how long each answer takes and how often no answer comes back. The result only applies to your connection: the fastest resolver for someone in London can be one of the slowest on your network.

A good benchmark reports more than a single number:

To understand why cached answers are faster, read what is DNS cache.

DNS benchmark tools compared

You will also see older names in search results. Namebench, the tool that started at Google, has not been maintained for years and its GitHub repository is archived. GRC no longer offers the free version 1 of DNS Benchmark: according to GRC it focused on cached lookups, so its ranking no longer fits today's internet. Avoid downloading old copies from third-party sites. If you run your own DNS server and want to know how many queries per second it can handle, use dnsperf from DNS-OARC; that is a load test, not a way to pick a resolver for your connection.

How to use GRC DNS Benchmark

DNS Benchmark by Gibson Research Corporation (GRC) is the best-known tool for the job. Version 2 tests resolvers over IPv4, IPv6, DNS over HTTPS and DNS over TLS side by side, and can build a custom list of the 50 fastest resolvers for your location out of nearly 5,000 public ones.

1. Buy and download DNS Benchmark from grc.com and start it.
2. On first use, read the Introduction tab. The list is filled with well-known public resolvers and the DNS servers your computer currently uses.
3. Close anything else that uses the network, such as downloads, updates and video calls.
4. Open the Nameservers tab and click Run Benchmark.
5. Watch the bar chart while it runs: red is cached, green is uncached and purple is dotcom.
6. When it finishes, open the Conclusions tab for a plain-language summary of what the results mean for you. The raw numbers are under Tabular Data.
7. To test more candidates, click Rebuild Custom List. DNS Benchmark then compiles the 50 resolvers that are fastest from your location. Compare them with Run Benchmark.

A coloured dot next to each resolver shows its status. Orange means the server returns its own page for names that do not exist instead of an error. Skip those resolvers, because applications then get a wrong answer.

Free alternative: dnseval

dnseval is part of dnsdiag, a free set of DNS tools by Babak Farrokhi. Prebuilt binaries are available for Windows, macOS and Linux, or install it with pip3 install dnsdiag (Python 3.11 or newer).

1. Create a text file servers.txt with one resolver per line, for example 1.1.1.1, 8.8.8.8, 9.9.9.9 and the one you use now.
2. Run dnseval -c 20 -f servers.txt ljpc-hosting.nl in a terminal or command prompt. The -c 20 option sends twenty queries per server.
3. Compare the avg, min, max, stddev and lost columns (milliseconds and percent) for each resolver.

By default dnseval warms up each resolver's cache first, so you mainly measure cached responses. Add --skip-warmup to turn that off. Test encrypted DNS with --tls (DNS over TLS) or --doh (DNS over HTTPS). To follow one resolver for longer, use dnsping -c 50 -s 9.9.9.9 ljpc-hosting.nl.

Quick test with dig or PowerShell

For a first impression you do not need extra software. Run each test a few times: the first query may be uncached, later ones usually come from the cache.

macOS and Linux: dig

dig is included in macOS Terminal. On Linux it comes with the dnsutils or bind9-dnsutils package (Debian, Ubuntu) or bind-utils (Fedora). This loop queries three resolvers and prints the Query time in milliseconds for each:

for s in 1.1.1.1 8.8.8.8 9.9.9.9; do printf "%s " $s; dig +noall +stats ljpc-hosting.nl @$s | grep "Query time"; done

Windows: PowerShell

Open PowerShell and run this command. It times one A-record lookup per resolver:

foreach ($s in '1.1.1.1','8.8.8.8','9.9.9.9') { $t = (Measure-Command { Resolve-DnsName ljpc-hosting.nl -Type A -Server $s -DnsOnly -NoHostsFile }).TotalMilliseconds; "$s $([math]::Round($t)) ms" }

The measured time includes PowerShell's own overhead, so only compare results you measured the same way.

Tips for accurate results

• Test where you will use the resolver. Run the benchmark on your own computer, at home or at the office. Rankings from other people say little about your connection.
• Keep the network quiet. Downloads, updates, streaming and video calls skew the results, especially the reliability figures.
• Turn off your VPN. With a VPN your DNS queries usually travel through the tunnel, so you would be measuring the VPN as well.
• Repeat the test at different times. Look at averages and outliers, not at one single run.
• Include your current resolver. That shows what switching would actually gain you.
• Test what you plan to use. If you want encrypted DNS or IPv6, benchmark over those as well.

How to read the results and choose a DNS server

1. Drop resolvers that lose queries. A resolver with lost queries is out, even if it is fast on average: after a missed answer your device waits for a timeout before it tries again.
2. Give uncached results real weight. A web page loads resources from many different domains, so how fast a resolver looks up names it has not seen before matters more often than you might expect.
3. Check the spread. A resolver with consistent response times feels smoother than one with large spikes.
4. Let other factors decide a close race. When the difference is a few milliseconds, choose on privacy, malware protection and encrypted DNS. The trade-offs are covered in public DNS vs ISP DNS servers.
5. Set up the winner. Enter two addresses from the same service, in your router or per device. The same article has the steps for Windows, macOS, iPhone, Android and routers.

Will a faster DNS server make your internet faster?

Only name lookups get faster. Your device keeps each answer in its own cache for as long as the record's TTL allows, so the gain is mostly on the first visit to a site. Your download speed stays the same, and how fast a website loads after the lookup depends on the server and the site itself. To speed up your own website, read speed up your website with caching and compression.

DNS performance of your own domain at LJPc hosting

A benchmark measures your visitors' resolvers. For your own domain it also matters how quickly its authoritative nameservers respond, because that is where every resolver fetches the answer when it is not cached. If your domain uses the LJPc nameservers, those are ns1, ns2 and ns3.ljpc.network, reachable over IPv4 and IPv6.

• Measure: dig yourdomain.com @ns1.ljpc.network shows under Query time how quickly our nameserver answers from your location, directly and without a resolver cache in between.
• TTL: the longer a record's TTL, the more often resolvers can answer from their cache. On our nameservers the TTL is usually one hour. Before a migration, a shorter TTL makes the change reach everyone sooner; see what is DNS propagation.
• Check the setup: Zonemaster, run by the .fr and .se registries, checks whether all nameservers of your domain are reachable, give the same answers and have DNSSEC set up correctly. It does not measure speed.

You manage your records in Plesk or in the customer portal at portal.ljpc.nl; see manage DNS records. Which nameservers belong to your domain is explained in NS record explained. Still seeing an old answer after a change, or is your domain not resolving? Troubleshoot DNS problems walks you through it. Need a hand? Create a support ticket and we will look into it with you.