Test the cloud platform of the Top 3
Hello, Habr! I want to share my results of a comparison of cloud platforms AWS, Azure and Google. Below is the performance results that gets a regular user, and their comparison by criterion "price/quality".
To begin, I introduced the figure as the number of points by means of testing, divided by the value of this virtual machine in an hour. To compare the performance of the virtual machines we used the following "instances":
the
* Current price for promotional positions in brackets regular price.
** Price with 30% discount if you use the virtual machines 24/7 in a month.
update: after publishing the article there was news about OS drives are smaller in Azure that allow you to save another $2.18 per month. The results of this does not change, but saves on the order of 0.8% of the price of a virtual machine (Standard D3 v2).
All "instances" have been "raised" in Europe – Ireland\Germany\Belgium. Creation took place through the appropriate portals of cloud providers. For instance SSD used in various configurations:
In order to make the tests as objective as possible, I did not set up the virtual machine. All "instances" use Windows Server 2016, and for the last test – 16.04.02 Ubuntu LTS. I also refused any settings of the OS and had to install the GeekBench test. I used the trial version from the websites of the manufacturer. Using apt-get for Ubuntu was installed benchmark Hardinfo. These are the only settings made with the operating system before running the tests.
the
On each virtual machine I 5 times consistently "chased" the following tests:
The most detailed description I found on the link. Briefly, the total number of tests – 23, they are divided into several categories: Cryptography, Integer, Floating Point, Memory. But in fact, the results of 25 tests, perhaps because I have a newer version. In the tests used compression operations, work with JPEG, HTML parsing, SQLite. I would call it one of the most comprehensive tests on different types of operations.
Here I couldn't find description, but as far as I understand, this program will run tests sequential write / read and random write\read + the same tests, but with queue depth equal to 32.
Just gonna leave this here: CPU Blowfish, CPU CryptoHash, CPU Fibonacci, CPU N-Queens, FPU FFT, and FPU Raytracing.
the
The results of the multi-core test the most interesting, if you look at the information about the tested systems, we can note that AWS and Google Cloud to give a picture of: Intel Xeon @ xxx GHz 1 processor, 2 cores, 4 threads, while the Azure gives "honest" core: Intel Xeon E5-2673 v3 @ xxx GHz 1 processor, 4 cores. I suspect that this is the reason almost polutorakratnoe the productivity gap. The higher the number in the test, the better.
the
total: Azure, AWS, Google Cloud.
More detailed results can be viewed in repository.
The results of CrystalDiskMark, the most ambiguous, but most likely this difference is due to the fact that the Google Cloud platform\AWS should further configured for optimal performance, while the Azure default configuration show good performance. In addition, I did not take into account the price, as it is difficult to calculate. The more numbers in the tests, the better.
In this configuration, Azure left the competition far behind on all indicators: sequential read / write, "random" 4K read / write, "random" 4K read\write with queue depth equal to 32. I think the result is interesting, because this machine by default, hence this result which will get the average user.
total: Azure, AWS, Google
Interpretation of test results Hardinfo I will leave readers. All test participants showed approximately the same result.
the
All the clouds have their own strengths and weaknesses:
In terms of power in monetary terms, the picture changes somewhat and Google with 30% discount begins to lose Azure, on average, ~10%, although when compared to the forehead calculators both platforms may seem quite the opposite. Amazon, unfortunately, begins to lose 28% in the money for the performance tests GeekBench.
Of course these are synthetic tests and for each individual solution or application you need to carry out specialized activity, but it can be a good starting base point in the selection.
Article based on information from habrahabr.ru
To begin, I introduced the figure as the number of points by means of testing, divided by the value of this virtual machine in an hour. To compare the performance of the virtual machines we used the following "instances":
| Cloud | Vm Size | Cores | Ram | Price\Hour on windows |
|---|---|---|---|---|
| AWS | m4.xlarge | 4 | 16GB | $0.406 |
| Azure | Standard D3 v2 | 4 | 14GB | $0.422 (0.488)* |
| n1-standard-4 | 4 | 15GB | $0.306** |
* Current price for promotional positions in brackets regular price.
** Price with 30% discount if you use the virtual machines 24/7 in a month.
update: after publishing the article there was news about OS drives are smaller in Azure that allow you to save another $2.18 per month. The results of this does not change, but saves on the order of 0.8% of the price of a virtual machine (Standard D3 v2).
All "instances" have been "raised" in Europe – Ireland\Germany\Belgium. Creation took place through the appropriate portals of cloud providers. For instance SSD used in various configurations:
the
- Google Cloud – 100GB, high for the "trial";
the- Amazon – 30GB, maximum, default;
the- Azure – 127GB by default.
In order to make the tests as objective as possible, I did not set up the virtual machine. All "instances" use Windows Server 2016, and for the last test – 16.04.02 Ubuntu LTS. I also refused any settings of the OS and had to install the GeekBench test. I used the trial version from the websites of the manufacturer. Using apt-get for Ubuntu was installed benchmark Hardinfo. These are the only settings made with the operating system before running the tests.
the
What tests were used
On each virtual machine I 5 times consistently "chased" the following tests:
the
- GeekBench (Windows only).
the- CrystalDiskMark (Windows only).
the- Hardinfo (only Ubuntu).
Update
The most detailed description I found on the link. Briefly, the total number of tests – 23, they are divided into several categories: Cryptography, Integer, Floating Point, Memory. But in fact, the results of 25 tests, perhaps because I have a newer version. In the tests used compression operations, work with JPEG, HTML parsing, SQLite. I would call it one of the most comprehensive tests on different types of operations.
CrystalDiskMark
Here I couldn't find description, but as far as I understand, this program will run tests sequential write / read and random write\read + the same tests, but with queue depth equal to 32.
Hardinfo
Just gonna leave this here: CPU Blowfish, CPU CryptoHash, CPU Fibonacci, CPU N-Queens, FPU FFT, and FPU Raytracing.
the
Results
Update
The results of the multi-core test the most interesting, if you look at the information about the tested systems, we can note that AWS and Google Cloud to give a picture of: Intel Xeon @ xxx GHz 1 processor, 2 cores, 4 threads, while the Azure gives "honest" core: Intel Xeon E5-2673 v3 @ xxx GHz 1 processor, 4 cores. I suspect that this is the reason almost polutorakratnoe the productivity gap. The higher the number in the test, the better.
| Cloud | GeekBench Score (1) | Price per hour | Perf Score / Price (2) |
|---|---|---|---|
| AWS | 6568.6 | $0.406 | 16 177 |
| Azure | 9508.4 | 22 530 | |
| 6188.2 | $0.306 | 20 222 |
total: Azure, AWS, Google Cloud.
More detailed results can be viewed in repository.
CrystalDiskMark
The results of CrystalDiskMark, the most ambiguous, but most likely this difference is due to the fact that the Google Cloud platform\AWS should further configured for optimal performance, while the Azure default configuration show good performance. In addition, I did not take into account the price, as it is difficult to calculate. The more numbers in the tests, the better.
In this configuration, Azure left the competition far behind on all indicators: sequential read / write, "random" 4K read / write, "random" 4K read\write with queue depth equal to 32. I think the result is interesting, because this machine by default, hence this result which will get the average user.
total: Azure, AWS, Google
Hardinfo
Interpretation of test results Hardinfo I will leave readers. All test participants showed approximately the same result.
the
Insights
All the clouds have their own strengths and weaknesses:
the
- Google – the best system of discounting, which is automatically activated and, at first glance, it really gives a big bonus.
the- Amazon – very strong from the point of view of confidence in the brand and the most popular platform.
the- Azurе has a performance advantage due to the sale of real cores, not virtual hyper-threading threads (vcpus).
In terms of power in monetary terms, the picture changes somewhat and Google with 30% discount begins to lose Azure, on average, ~10%, although when compared to the forehead calculators both platforms may seem quite the opposite. Amazon, unfortunately, begins to lose 28% in the money for the performance tests GeekBench.
Of course these are synthetic tests and for each individual solution or application you need to carry out specialized activity, but it can be a good starting base point in the selection.
Комментарии
Отправить комментарий