All grades are based on tests run by VPSBenchmarks on each VPS plan.
There are 7 types of tests:
web, sysbench, endurance, geekbench, fio, iperf3 and network transfers.
Each grade is the combination of one or more test metrics.
A grade is calculated in each on of 5 categories: web, cpu, disk, network and stability.
Each category grade is made of one or more metrics chosen in relevant test types.
Given a metric for a plan and a test, the grade for that metric is determined relative to the same metric from all other plans.
A plan with a metric in the top 18% of all plans will get an A,
a plan with a metric in the bottom 18%
will get an F. Relevant single metric grades are then weighted and combined
to calculate a category grade.
Under the hood, a grade is a decimal number between 0.0 and 6.0.
The overall score is the average of the grades in all categories
brought into the [0 - 100] range.
Note: The VPS plan price is not a factor in calculating the
grades. More expensive plans with higher specs are likely to get higher grades
than cheaper ones. This can be changed in the Screener by turning on the "price weighted" switch.
Grade Categories
Web Performance
The Web Performance grade is made of the Web test average response time,
99th percentile response time and the maximum rate of requests without error.
Raw CPU Power
The Raw CPU Capacity grade is made of the Sysbench and Geekbench CPU test
number of operations per second and of the Endurance test
number of iterations per hour metric.
Performance Stability
The Performance Stability grade is made of the Web test average response time
standard deviation, the difference between the 99th percentile and average response times
and the Endurance test number of iterations per hour standard deviation.
Disk IO Performance
The Disk IO Performance grade is made of the Sysbench and Fio tests measuring random and sequential storage speeds.
Network Performance
The Network Performance grade is made of Speedtest and Iperf3
network upload and download transfer speeds.
Price Weighted Grades
The Screener and other pages offer the option to display "price weighted" grades.
With this option turned on, a low price plan that has the same performance as
a high price plan will be shown with higher grades.
The price weighted view tries to answer the following questions:
What plans have the best value?
How much performance do I get for every dollar I spend?
For low price plans, the grade "bonus" is proportional to the difference between its $30 and monthly price.
The function used to calculate the bonus is designed
to have a small impact on very low grades and maximum impact on medium to high grades.
Similarly, grades for high price plans are displayed lower when this option is turned on:
the higher above $30 the monthly price is, the stronger the hit on the grades.
In this case, the function is designed to have a small impact on excellent grades
and maximum impact on medium to low grades.
Performance Grades
To summarize the performance of each VPS plan, we generated grades based on the various tests that were run:
web, sysbench, endurance and remote timing tests. Each grade focuses on one aspect of the VPS speed or stability
(cpu, disk IO, network IO...) and is made of one or more metrics measured during the tests.
Grades are evenly distributed between A and F. A is best, F is worst.
Follow the letter links to find the test results backing the grade.
The provider consistency score shows how much variation can be expected between 2 servers of the same type at the same provider. A high consistency score (>65) means the performance of all servers of the provider is consistent over many trials and many plans.
Providers who don't have enough data yet at VPSBenchmarks don't have a consistency score.
These are some of the test results that are used to determine the grades above.
Web Response Times
We ran a database intensive web application at various levels of load on cloud servers from
ORACLE Cloud. This is the average response time measured locally on the VPS for each plan we tested.
Local Response Time per Plan
Lower is better
Maximum HTTP Load
We ran a database intensive web application at various levels of load on cloud servers from
ORACLE Cloud.
This is the maximum rate of HTTP requests that each server handled without timing out.
Maximum HTTP request rate
Higher is better
During the Web Test, we collect response time metrics, rate metrics and CPU metrics.
All of those are measured locally on the VPS running the web server.
All metrics can be compared on the Web Performance
Comparison page.
We ran the full suite of Sysbench tests on VPS plans of ORACLE Cloud.
These are the Sysbench CPU test results:
Rate of operations
Higher is better
Sysbench Disk IO Performance
We ran the full suite of Sysbench tests on VPS plans of ORACLE Cloud.
These are the Sysbench Disk IO Random Read test results:
Rate of operations
Higher is better
Sysbench test results are also available for random disk IO tests,
sequential disk IO tests and memory tests.
Review and compare all Sysbench numbers on the Sysbench Comparison page
for ORACLE Cloud.
We have performed large file transfers from the ORACLE Cloud VPS and we measured the speed of downloads and uploads.
The transfers involve multiple files, 10 threads and last for at least 10 seconds at maximum speed. The transfers are repeated 3 times from different servers and only the fastest transfer is reported here.
Download Speed per Plan
Higher is better
Network Transfers measurements including download and upload speeds can be compared for
ORACLE Cloud
on the Network Transfers Comparison page.
We pushed the CPU of those VPS plans hard for 24 hours straight.
This test shows how fast the VPS can go for long periods of time, how stable
the performance is and it also gives clues on how overallocated the VPSs may be.
Throughput variations of the endurance test over 24 hours.
The coefficient of variation is the standard deviation of the test throughput divided by its average value.
Low variations mean that resources allocated to the server were stable for the duration of the test and there was little contention.
Ops/hour coefficient of variation
Lower is better
Review and compare endurance test results for ORACLE Cloud
on the Sustained CPU Endurance Comparison page.
With "Private Trials", the same benchmarks that run on all VPS at vpsbenchmarks.com will run on the servers of your choice.
The results are private and you will be able to compare them to any other public benchmark
on this website.
We are proud to provide objective and impartial benchmark data on this website.
VPSBenchmarks receives support from some of the providers featured here but all tests are
conducted the same way regardless of our relationships with them.
If you find our benchmarks valuable, you can help by making your cloud purchases using the
Provider Affiliate Buttons
displayed throughout the site.