Hetzner Uptime and Speed Test: How to Check Both

Hetzner Uptime and Speed Test: How to Check Both

Hetzner's uptime and speed can be checked in three complementary ways: an official status page for service incidents, per-location ping hosts and test files for latency and throughput, and independent monitors for a second opinion. No single test proves overall reliability, so the useful approach is to combine all three and read them against your own location and workload.

Uptime, Latency, and Throughput: Three Different Measurements

These three terms are often used interchangeably, but they answer different questions.

Uptime is whether a service is reachable at all. It is measured over time, usually as a percentage of minutes or hours in a month when the service responded.

Latency is the time a system takes to respond after a request is made, measured in milliseconds or seconds. It determines how quickly a page starts loading or how a remote command feels in a terminal.

Throughput is the quantity of requests a system can process within an interval, measured in requests per second, transactions per second, or bits per second. It determines how fast a large download or backup completes.

Low latency matters most for real-time responsiveness, while high throughput matters most for handling more load. The two can pull in opposite directions: as throughput increases, latency often rises, indicating performance strain. That is why a good Hetzner evaluation looks at both rather than treating one number as the verdict.

Checking Uptime with the Status Page and Independent Monitors

Hetzner maintains an official status page for its services. Third-party aggregators track and log the outages and problems reported there, which makes it easier to look back at incident history rather than only the current state.

Independent monitors add a second layer. One such monitor probes Hetzner from six regions on every continent, checking every minute, and publishes 30-day uptime figures for individual services and locations. In one snapshot, it reported 99.99% uptime for the Hetzner website and 99.98% for Falkenstein (FSN1) over 30 days.

Treat the status page as the primary record of provider-side incidents and independent monitors as a cross-check. A third-party figure is a snapshot from that tool's probes, not a contractual uptime guarantee, and the monitor's own methodology and probe details are not fully documented. The two sources together help you separate a Hetzner-wide outage from a problem that is local to your server or application.

Testing Latency and Speed with per-Location Hosts and Test Files

Hetzner runs data center locations in Falkenstein and Nuremberg in Germany, Helsinki in Finland, Ashburn and Hillsboro in the United States, and Singapore. For each location, it provides a ping host you can use to check latency, plus a speed test page with 100 MB, 1 GB, and 10 GB files to download.

The ping hosts follow a consistent naming pattern: hil.icmp.hetzner.com, ash.icmp.hetzner.com, fsn.icmp.hetzner.com, nbg.icmp.hetzner.com, hel.icmp.hetzner.com, and sin.icmp.hetzner.com. Running a standard ping against the host for a location gives you a rough round-trip time from your network to that site. Hetzner does not provide a looking glass tool for testing a server, so ping and file downloads are the main first-party options.

For browser-based checks before you deploy anything, third-party tools fill the gap. One Hetzner ping test measures latency from your browser to locations in Falkenstein, Nuremberg, Helsinki, Ashburn, and Hillsboro. Another browser-based latency test measures by making a small HTTP request and estimating TLS, TCP, and HTTP overhead, updating every 2 seconds. These are convenient for a quick comparison of which region answers fastest from where you sit. They measure latency only, and the browser-based estimate includes protocol overhead, so the number will not match a raw ICMP ping exactly.

The location guide that documents these per-location hosts and test files also covers how Hetzner groups its cloud locations, which is worth reading alongside the test results: Hetzner data center locations.

What the Benchmark Evidence Shows

For throughput, independent benchmarking gives a more comparable picture than a single download. In a benchmark of 4 vCPU / 8 GB shared VPS plans, Hetzner at $18.59/mo recorded a median network download of 1,811 Mbps and upload of 1,689 Mbps. The benchmark ran roughly 1,200 automated tests per server using sysbench, fio, and speedtest-cli, and it placed Hetzner as the only provider in the top-right quadrant for strong CPU and strong network. You can review the full VPS benchmark for the surrounding results.

Keep the scope in mind: this covers one plan tier at one point in time, so it should not be generalized to every Hetzner plan or location. The general relationship between the two metrics is explained in this latency versus throughput guide.

There is a legitimate debate about provider-hosted tests. TestMy.net argues that third-party speed tests are preferable because provider-hosted tests eliminate the variables users want to test. The practical resolution is to use Hetzner's own test files for a consistent baseline from each location, and third-party tools to confirm what a path looks like from outside Hetzner's network.

Running a Fair, Repeatable Test

A single speed test is a data point, not a conclusion. To get results you can actually compare:

  1. Test from the network your users will actually use, not from a different connection.
  2. Run each test at least three times and note the median, not the best result.
  3. Use the same file size across locations so the comparison is like-for-like.
  4. Test at different times of day to catch congestion effects.
  5. Record the location, date, time, file size, and result in a simple log.

Run latency tests with ping to the per-location hosts and throughput tests by downloading the 100 MB, 1 GB, or 10 GB files. Start with the small file for a quick check and use the larger files when you want a steadier throughput figure. Browser-based latency tools are fine for a fast regional comparison, but keep them separate from your ping and download records because they measure differently.

Reading the Results: Distance, Network Zones, Plan Tier, and Time of Day

A raw number means little without context. Four factors explain most of the variation you will see.

Distance. Physical distance between your users and the data center sets a floor on latency. Hetzner's German locations are described as offering strong connectivity across Europe, with Nuremberg and Falkenstein typically delivering low latency for European audiences. The US locations in Ashburn and Hillsboro extend that reach to North American users.

Network zones. Hetzner Cloud locations are grouped into four network zones: eu-central (Falkenstein, Nuremberg, Helsinki), us-east (Ashburn), us-west (Hillsboro), and ap-southeast (Singapore). This matters for more than latency: private networks, Floating IPs, and Load Balancers only work between servers in the same network zone. Traffic between two zones is billed like normal internet traffic. A fast result in a location you cannot privately network with your other servers may not help your architecture.

Plan tier and product availability. Not every product is sold in every location. Shared AMD and dedicated AMD cloud servers are available in all six locations, but Arm-based Ampere servers, the shared Intel/AMD line, Object Storage, and dedicated root servers are only offered in Germany and Finland. A location that measures well is only useful if the plan type you need is sold there.

Time of day and congestion. Latency and throughput vary with load on your own connection, the transit path, and the destination. This is a general networking effect rather than a Hetzner-specific measurement, so repeated tests at different hours are the only reliable way to see it in your own results.

Facility differences are also worth knowing. Hetzner's own German and Finnish data center parks use redundant UPS (N+1), a 2.5 MVA diesel generator, dual independent power lines, outside-air cooling with N+2 redundancy, 24/7 bilingual on-site service, video surveillance, DIN ISO/IEC 27001:2022 certification, DDoS protection, and air-sampling smoke detectors. In the US and Singapore, the building, power, and cooling belong to the colocation provider rather than Hetzner. Hetzner says its DDoS protection works the same way in Ashburn as in its other data centers. These are infrastructure facts rather than speed measurements, but they explain why two locations can behave differently under stress.

Matching Results to Your Audience and Workload

There is no single best Hetzner location. The right choice depends on where your users are and what you are running.

  • European audience, full product range needed. Falkenstein, Nuremberg, or Helsinki are the natural fit: strong European connectivity and access to Arm Ampere, shared Intel/AMD, Object Storage, and dedicated root servers. All three European locations share the same pricing. One independent monitor measured slightly higher latency for Helsinki than the German sites, so compare the three with your own ping tests before deciding.
  • North American audience. Ashburn serves the eastern US and Hillsboro the western US. Both support shared AMD and dedicated AMD cloud servers, but not the Arm, Intel/AMD shared, Object Storage, or dedicated root lines. Prices are higher than in Europe because operating costs are higher there.
  • Asian or Australian audience. Singapore is the closest Hetzner location and supports shared AMD and dedicated AMD cloud servers, with the same narrower product range and higher pricing as the US sites.

A practical decision rule: ping each candidate location from your own network, download a test file from each to compare throughput, confirm the plan type you need is sold there, then check the status page and an independent monitor for that location's recent reliability. That sequence uses each test for what it actually measures.

Limitations and What Still Needs Verification

Some questions cannot be answered from the available evidence and are worth checking directly before you commit.

  • Hetzner's official status page is the authoritative incident record, but the figures here come from third-party monitors and should be read as a snapshot, not as an official uptime history.
  • Exact current URLs and file sizes for the per-location test pages are reported by third parties; confirm them on Hetzner's own pages before relying on a specific link.
  • The throughput benchmark covers one shared 4 vCPU / 8 GB plan at one point in time, not all plans or locations.
  • No per-location throughput results from Hetzner's own test files are available here, so location-by-location speed comparisons need your own downloads.
  • Contractual uptime guarantees and SLA credit terms are not covered by these sources. If a service-level agreement is part of your decision, read Hetzner's own terms.
  • Dedicated server network performance is not covered separately from cloud VPS performance.
  • How DDoS protection affects measured latency or throughput during an attack is not documented in these sources.

Used together, the status page, the per-location ping hosts and test files, and independent benchmarks give you a defensible picture of Hetzner's uptime and speed before you buy. The numbers are most useful when you generate some of them yourself, from your own network, against the locations and plans you are actually considering.

Leave a Comment

Your email address will not be published. Required fields are marked *

Scroll to Top