Run an authorized, controlled test with explicit request and concurrency counts, then judge throughput, latency, and correctness together. A useful starting point is ab -n 1000 -c 10 https://your-vps.example/. Repeat it at gradually higher concurrency, save the complete output, and record the VPS and endpoint conditions so the results can be reproduced.
Antihost offers Linux VPS plans with full root access, unlimited traffic on a 1 Gbps port and a 14-day free trial.
What ApacheBench measures
ApacheBench, whose executable is ab, is Apache HTTP Server’s focused HTTP benchmarking utility. Apache describes it as a tool for benchmarking an Apache HyperText Transfer Protocol (HTTP) server and uses it to provide an impression of current server performance, particularly requests per second.
As an Amazon Associate I earn from qualifying purchases.
The Tool Desk
Outbyte PC Repair FREERepair Windows errors before they cause bigger problemsFix Now →Outbyte Driver Updater FREEFix the driver behind crashes, sound loss and screen glitchesFind Drivers →The utility generates a controlled stream of HTTP/1.x requests against one endpoint. It is useful for comparing a configuration with itself, checking whether a change altered performance, and finding the point at which an endpoint stops responding cleanly. It is not a complete simulation of a browser, a modern single-page application, or every user journey.
Do not rely on the default one-request invocation. Apache’s documentation warns that a single request normally produces a non-representative result. Always specify both -n, the total number of requests, and -c, the number of clients running concurrently.
#1 Best Overall
- Read Before You Buy — No Video Output: These adapters support charging and USB 2.0 data transfer, but cannot transmit video signals. Except for standard USB webcams (which use USB data only), they are not compatible with HDMI/DisplayPort cables, video-capable USB-C hubs, or docking stations with video output.
- Convert USB-A Ports to USB-C: Designed to connect USB-C earphones, cables, flash drives, card readers, and other USB-C accessories to standard USB-A ports. Plug-and-play with no drivers or software required.
- Aluminum Alloy Housing: Built with a sturdy aluminum alloy shell that aids in heat dissipation and protects against daily wear and scratches. Designed to maintain a stable and secure connection.
- Compact & Travel-Friendly: The ultra-compact design allows the adapter to stay plugged into your device without blocking adjacent ports or adding bulk, reducing wear and tear on your original USB ports.
- 12-Month Warranty: Backed by a 12-month manufacturer warranty for peace of mind. Designed to meet strict quality control standards for reliable everyday performance.
Before you test: authorization and a repeatable target
Benchmark only a VPS and endpoint that you own or are explicitly authorized to test. A burst of new connections can resemble a denial-of-service event and can activate protective controls on the server, host, firewall, or network.
Choose one endpoint
Use a URL whose behavior you understand and keep it fixed while comparing runs. Record the following before the first request:
- The complete URL, including the path and query string.
- The HTTP method and request body, if the endpoint requires something other than a simple request.
- Whether authentication, cookies, or a particular account state is involved.
- The response status and approximate response size.
- Whether the request uses HTTP or HTTPS and, for HTTPS, the TLS configuration in use.
- Whether persistent connections (keep-alive) match the production traffic you are trying to understand.
A static health endpoint, a cached page, and a database-backed dynamic page exercise very different parts of a VPS. A result from one cannot be treated as a capacity figure for the others.
Record the test context
For every run, note the date and time, VPS plan or size, virtual CPU and memory allocation, storage activity, network path, web-server configuration, application version, cache state, and any background jobs. Also record the exact ab command. A result without these conditions is difficult to reproduce and unsafe to compare with another result.
Install the ab executable
The package name depends on the Linux distribution. Debian- and Ubuntu-based systems commonly provide the executable in apache2-utils; Red Hat-family distributions commonly use httpd-tools. Use the package manager’s normal repository for your distribution.
Debian or Ubuntu example
sudo apt update
sudo apt install apache2-utilsRHEL, Fedora, or compatible systems
sudo dnf install httpd-toolsOn an older system that still uses YUM, use its equivalent package-install command. Confirm that the executable is available before testing:
ab -VThe version output is useful context to save with your results. Package names and available command-line options can vary by distribution and Apache version.
Rank #2
- 5-in-1 USB-C Hub: Experience comprehensive connectivity featuring a Power Delivery input, two USB-A 2.0 ports, a USB-A 3.0 port, and an HDMI port. (Note: The USB-C power delivery input port is only for connecting an external wall charger to power your laptop and cannot power peripheral devices.)
- 90W Pass-Through Charging: Achieve optimal charging with 90W pass-through power to your laptop, supported by a total input of 100W, with the hub reserving 10W for operational efficiency. (Note: Wall charger not included.)
- Quick Data Transfers: Accelerate your productivity with rapid data transfers using a high-speed 5Gbps USB 3.0 port and two 480Mbps USB 2.0 ports.
- 4K HDMI Display: Enhance your visual experience with a hub capable of delivering 4K resolution at 30Hz in both mirror and extend modes. Please note that this hub is compatible with MacBook (macOS 12 and newer), Windows 10 and 11, ChromeOS, and laptops equipped with DP Alt Mode and Power Delivery. Note: This device is not compatible with Linux.
- What You Get: Anker USB-C Hub (5-in-1, 4K HDMI), welcome guide, 18-month warranty, and our friendly customer service.
Run a conservative baseline
Start with a modest, explicit workload:
ab -n 1000 -c 10 https://your-vps.example/Here, -n 1000 asks for 1,000 total requests and -c 10 allows up to 10 requests to be active at once. The command should target the endpoint you recorded, not an arbitrary public site.
- Confirm that the endpoint is healthy with a normal request before starting the benchmark.
- Run the command from a machine whose network path and location you can document. The client’s distance, bandwidth, and packet loss affect the measured time.
- Wait for the run to finish without changing the server configuration or starting unrelated load.
- Save all terminal output, including warnings and the final summary.
- Check the server and application logs for errors, rate limits, restarts, and unusual resource pressure.
The baseline is a reference point, not a promised performance target. Its purpose is to establish how the endpoint behaves under a known, repeatable workload.
Increase concurrency in steps
After the baseline, repeat the same request count and endpoint at several conservative concurrency levels, such as 1, 5, 10, and 25:
Quick wins for a faster PC:
Clear out junk files and repair common Windows errorsFree Scan →Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →ab -n 1000 -c 1 https://your-vps.example/
ab -n 1000 -c 5 https://your-vps.example/
ab -n 1000 -c 10 https://your-vps.example/
ab -n 1000 -c 25 https://your-vps.example/Run each level only while the service remains healthy and the test remains authorized. A higher concurrency value is not automatically better: it may expose useful saturation behavior, or it may simply overload the endpoint and invalidate the measurement.
Keep -n constant while changing -c if you want to isolate the effect of concurrency. If you change both values, label the runs as different workloads rather than treating them as a simple progression.
Options that change what the test means
| Option | Meaning | When to use it | Important qualification |
|---|---|---|---|
-n NUMBER |
Total number of requests to perform | Every benchmark, so the sample size is explicit | The one-request default is usually not representative |
-c NUMBER |
Number of concurrent clients | To test a defined level of parallel demand | Increase gradually; excessive bursts can trigger protective controls |
-k |
Use HTTP keep-alive connections | When persistent connections are part of the production pattern | Compare keep-alive and non-keep-alive runs only as separate workloads |
-s SECONDS |
Bound the socket wait time | To prevent a stalled connection from waiting indefinitely and to make a timeout policy explicit | A timeout changes the failure behavior; report the value with the result |
-e FILE.csv |
Export percentile timing data | When average latency is not enough and tail behavior matters | The file contains timings for each percentage from 1% through 100% |
-R MILLISECONDS |
Delay between connection starts | When you need a gentler ramp instead of an immediate connection burst | Available in versions that support it; the run can finish before full concurrency is reached |
For example, a keep-alive run with a bounded wait and percentile export could be written as:
ab -n 1000 -c 10 -k -s 30 -e results.csv https://your-vps.example/Use options to answer a specific question. Do not add every flag by habit: a test that does not match the traffic pattern you care about can be precise yet irrelevant.
Recommended Free Tools
Rank #3
- Sleek 7-in-1 USB-C Hub: Features an HDMI port, two USB-A 3.0 ports, and a USB-C data port, each providing 5Gbps transfer speeds. It also includes a USB-C PD input port for charging up to 100W and dual SD and TF card slots, all in a compact design.
- Flawless 4K@60Hz Video with HDMI: Delivers exceptional clarity and smoothness with its 4K@60Hz HDMI port, making it ideal for high-definition presentations and entertainment. (Note: Only the HDMI port supports video projection; the USB-C port is for data transfer only.)
- Double Up on Efficiency: The two USB-A 3.0 ports and a USB-C port support a fast 5Gbps data rate, significantly boosting your transfer speeds and improving productivity.
- Fast and Reliable 85W Charging: Offers high-capacity, speedy charging for laptops up to 85W, so you spend less time tethered to an outlet and more time being productive.
- What You Get: Anker USB-C Hub (7-in-1), welcome guide, 18-month warranty, and our friendly customer service.
How to read the summary
ApacheBench prints a summary after the requests complete. The following fields are the ones needed for a capacity decision.
| Field | ApacheBench definition | How to use it |
|---|---|---|
| Concurrency Level | The number of concurrent clients | Verify that the run used the concurrency you intended. If ramping was used, check the achieved value rather than assuming the requested value was reached. |
| Complete requests | Requests that completed successfully | Use it with the total request count to confirm how much of the workload finished normally. |
| Failed requests | Requests ApacheBench considered failures | Any increase is a correctness or availability problem, even if throughput also rises. |
| Non-2xx responses | Responses outside the HTTP 200 series | Investigate redirects, authentication responses, client errors, server errors, and rate limits before calling the run successful. |
| Requests per second | Requests divided by total elapsed time | A throughput indicator for this endpoint, client, network path, and workload—not a universal VPS rating. |
| Time per request | Average request time | Read it alongside concurrency, failures, status codes, and percentiles. An average can conceal a long tail. |
| Transfer rate | Bytes received per unit of time | Helps show whether the test is moving substantial response data and whether network transfer, rather than application work, may be influencing the result. |
Requests per second and time per request answer different questions. Throughput describes how much work completed over the whole run; latency describes how long requests took on average. Neither proves that every response was correct.
Judge throughput, latency, and correctness together
A run with more requests per second is not a capacity improvement if failed requests or non-2xx responses rise. Treat the three dimensions as a gate:
- Correctness: complete requests should account for the intended workload, with failed requests investigated rather than ignored.
- Status: non-2xx responses must be explained. A redirect or an expected authorization response may be intentional, but it should not be silently counted as a successful application transaction.
- Latency: average time should remain within the service’s acceptable range, and tail latency should not deteriorate sharply as concurrency increases.
- Throughput: requests per second should be compared only between runs with the same endpoint, request count, client path, and relevant connection behavior.
There is no universal Linux VPS requests-per-second figure established by Apache’s documentation. A number without the endpoint, response size, concurrency, keep-alive setting, client location, and server condition is not a meaningful promise about another VPS.
Use percentile data to expose tail latency
Average time can look acceptable while a minority of requests take far longer. Add -e when you need that distribution:
ab -n 5000 -c 25 -e results.csv https://your-vps.example/Apache documents that the exported file contains timings for every percentage from 1% through 100%. The high-percentile end shows how slow the slowest portion of the run became. Compare the shape of the distribution as concurrency rises: a sharp worsening near the upper percentages is evidence that some clients are waiting on a bottleneck even if the average still appears stable.
Keep the CSV with the terminal summary and the test context. A percentile value is meaningful only when its workload and conditions are known.
Outdated Drivers Are Slowing You Down
One free scan finds every outdated or missing driver and matches the right update for your exact hardware.Free scan · exact hardware matchWindows Errors? Fix Them Before They Spread
Repair common Windows errors and clear accumulated junk for a smoother, more stable PC - no reinstall needed.Free scan · no reinstallRank #4
- Dual Converters, Infinite Potential：Includes 2× USB C male to USB A female adapters and 2× USB A male to USB C female adapters. Perfect for a wide range of uses—tablets with Bluetooth keyboards, expand USB ports on macbook, and more. Two different converters for all your daily needs
- Next-Level 10Gbps & 3A Charging: No more slow 480Mbps, this usb to usb c adapter has a transfer speed of up to 10Gbps, allowing you to do more transferring in less time. This usb adapter fits both USB A and USB C charger, supporting up to 3A fast charging
- Upgraded Exquisite Craftsmanship: With an aluminum alloy housing and metal connector, the usbc to usb adapter is extremely durable and sturdy. Rigorously tested to withstand more than 10,000 times of plugging and unplugging, ensuring long-lasting performance
- Broad Compatible: The usb c to usb adapter widely supports all USB C/ USB A devices like laptops, tablets, cellphones, car chargers, and phone chargers. Such as compatible with MacBook Pro/Air 2023/2022, Thunderbolt 4/3 Devices,Apple MagSafe Watch 9/8/7/SE/Ultra, iPad Pro 2022/2021, Samsung Galaxy S23/S20/S10, and iPhone 17/16/15 Pro. Plug and play
- Please Note: To reach 10Gbps speed, keep the cable under 3.3 ft. For USB A Male to USB C adapters, try flipping the USB C connector. USB C Male to USB A adapters support bidirectional 10Gbps transfer within 3.3 ft
Ramp connection starts when bursts are unsafe
By default, a high requested concurrency can start many connections in a short burst. Apache warns that starting too many concurrent connections at once can trigger denial-of-service protection. On versions that support it, -R MILLISECONDS inserts a delay between connection starts:
ab -n 1000 -c 25 -R 100 https://your-vps.example/Ramping reduces the abruptness of the arrival pattern, but it changes the workload. Apache also cautions that the test may finish before the requested concurrency is fully reached when connection starts are delayed. Inspect the reported achieved concurrency and record the ramp value; do not label the result as a full-concurrency run unless the output confirms that it reached that level.
Monitor the VPS while ab runs
ApacheBench’s summary is only the client-side view. Watch the host and application during every meaningful run and note which resource becomes constrained first.
- CPU pressure: sustained saturation can indicate application, TLS, compression, or web-server work that scales with each request.
- Memory pressure: swapping or reclaim activity can produce abrupt latency increases and failures.
- Storage activity: a dynamic endpoint may wait on disk or database I/O even when CPU is available.
- Network behavior: bandwidth limits, packet loss, and the path between the benchmark client and VPS affect transfer rate and timing.
- Service logs: look for upstream errors, worker exhaustion, rate limiting, restarts, and authentication failures.
Capture the monitoring context with the command output. If a run has an attractive throughput number but the VPS is swapping or the application is returning errors, treat it as an overload signal, not as usable capacity.
Keep comparisons scientifically fair
When evaluating a configuration change, alter one material factor at a time. Keep the endpoint, request count, concurrency ladder, keep-alive choice, timeout, ramp setting, client location, and software version fixed unless the purpose of the experiment is to test one of those variables.
Run more than one iteration when the environment is shared or variable, but do not average away failures. Preserve each complete output so an unusual run can be explained later. Avoid comparing a warm cache with a cold cache, a quiet VPS with one running backups, or an HTTP request with an HTTPS request and then attributing every difference to the web-server setting.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.Troubleshoot common results
Failed requests increase
Stop increasing concurrency. Check the full output and server logs, then determine whether the cause is a connection failure, timeout, resource exhaustion, application exception, or protective limit. Repeat only after the endpoint is healthy and the cause is understood.
Best Value
- 5-in-1 Connectivity: Equipped with a 4K HDMI port, a 5 Gbps USB-C data port, two 5 Gbps USB-A ports, and a USB C 100W PD-IN port. Note: The USB C 100W PD-IN port supports only charging and does not support data transfer devices such as headphones or speakers.
- Powerful Pass-Through Charging: Supports up to 85W pass-through charging so you can power up your laptop while you use the hub. Note: Pass-through charging requires a charger (not included). Note: To achieve full power for iPad, we recommend using a 45W wall charger.
- Transfer Files in Seconds: Move files to and from your laptop at speeds of up to 5 Gbps via the USB-C and USB-A data ports. Note: The USB C 5Gbps Data port does not support video output.
- HD Display: Connect to the HDMI port to stream or mirror content to an external monitor in resolutions of up to 4K@30Hz. Note: The USB-C ports do not support video output.
- What You Get: Anker 332 USB-C Hub (5-in-1), welcome guide, our worry-free 18-month warranty, and friendly customer service.
Non-2xx responses appear
Inspect the returned status and request assumptions. Authentication state, redirects, rate limiting, a missing host configuration, and application errors can all produce a response that is technically complete but not the success condition you intended to measure.
Do these 3 things before closing this tab:
1Fix the driver behind crashes, sound loss and screen glitches2Clear out junk files and repair common Windows errors3Scan for outdated or missing drivers - takes under a minuteRequests per second rises while latency worsens
Higher concurrency can increase total completed work while making individual requests wait longer. Use the percentile CSV and host monitoring to find the point at which the trade-off stops meeting your latency requirement.
The run times out or hangs
Set an explicit -s value appropriate to the service and record it. A timeout prevents an indefinitely stalled socket from obscuring the test, but timed-out requests are failures to investigate, not successful capacity.
The achieved concurrency is lower than requested
This can occur when connection starts are ramped with -R and the finite request set completes before the full level is reached. Check the reported concurrency, increase the request count only within an authorized and safe test plan, or remove the ramp only if an immediate burst is acceptable.
Results change dramatically between identical commands
Look for changes in cache state, background work, CPU or memory pressure, storage activity, network path, TLS behavior, and application logs. If the conditions are materially different, treat the runs as separate experiments rather than averaging them.
Free tools Windows power users keep installed
One-click scans. No signup required.
Turn the results into a capacity decision
- Define the endpoint and the acceptable status, failure, and latency conditions before testing.
- Establish a baseline with explicit
-nand-cvalues. - Repeat at progressively higher concurrency while observing the VPS and application.
- Mark the first level where failures, non-2xx responses, or unacceptable tail latency appear.
- Use the highest level that remains correct and within your latency objective as a test-specific operating point, not as a universal VPS rating.
- Repeat after configuration or code changes using the same documented workload.
This approach gives you an evidence-based boundary for one endpoint under one set of conditions. It does not predict every workload your site may receive.
Understand ApacheBench’s limits
ab is a focused HTTP/1.x request generator. It does not reproduce every browser workflow, cache pattern, connection mix, JavaScript action, authentication refresh, or multi-step application journey. A single endpoint can also hide contention between pages, APIs, queues, and background jobs.
Use ApacheBench for controlled endpoint checks and pair it with application and host monitoring when making capacity decisions. For a broader production model, you would need a workload that reflects the actual routes, request mix, session behavior, and traffic distribution; the ab results remain one controlled measurement within that larger assessment.
Quick Recap
Practical run checklist
- Authorization for the VPS and endpoint is confirmed.
- The URL, method, response size, authentication state, TLS use, and keep-alive expectation are recorded.
- The distribution package supplying
abis installed and its version is saved. - The command includes explicit
-nand-cvalues. - Concurrency levels are increased gradually, with protective limits and service health monitored.
-k,-s,-e, and, where supported,-Rare used only when they match the question being tested.- Complete requests, failed requests, non-2xx responses, requests per second, time per request, transfer rate, and achieved concurrency are reviewed together.
- Percentile output is retained when tail latency matters.
- VPS resource pressure, storage, network path, logs, date, and configuration are saved with every result.
- No result is presented as a universal requests-per-second rating for all Linux VPS instances.
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
