Hyperscale is a new trend when it comes to cloud technologies as hyperscalers continue to expand in the Middle East countries. We've seen AWS aiming at Bahrain and facing constant delays & Microsoft Azure which already setup a presence in the UAE.
We have decided to take a deeper look at what our competition has to offer and create a fair comparative study to breakdown each provider by quality, performance & price.
What is a hyperscale and who are the so-called hyperscalers?
The word itself defines it. The prefix “hyper” means over or excessive while the word scale refers to size or extent. So companies now being called “hyperscalers” are those massive companies like Microsoft and AWS that are making efforts to not only dominate the public cloud and cloud services industries but to expand their business into numerous related verticals.
As a small provider, we wanted to see how we compare when it comes to performance and the results were unbelievable. How? Let's find out.
The Test Servers
AEserver Cloud 4 Instance
We have selected the one of the most popular among our Cloud offers - Cloud 4 and we offer this for AED 289.00/month:
2 vCPU + 4GB RAM
60GB Disk Space (SSD)
Unmetered traffic @ 5mbps
Dubai Datacenter
Check out the plan HERE.
Amazon's - AWS c5.large Instance
We have selected c5.large instance inside AWS, with standard SSD storage as this instance is the closest to our configuration. It has 4GB RAM and 2 CPU too.
Microsoft's - Azure Standard B2s Instance
We have chosen a ‘Standard B2s’ which has the same specification as the servers above. 4GB RAM and 2 CPU cores available in this instance.
What to test?
To the actual performance of the machines, we have done 2 tests. First, the overall performance under production workloads using Unix Bench. UnixBench can be used to evaluate the performance of your system when running single or multiple tasks. It provides a simple scoring system for comparison of each instance.
The second one will be multiple runs of FIO to test the IO Performance of each machine. We will run a total of 9 Tests on each machine which will be the basis of the overall result.
The Preparation:
We have prepared CentOS 7 x64 installed with all updates completed in all of the servers to avoid any optimized version.
We are using KVM as the underlying technology. AWS is using XEN but moving to KVM as well in their public clouds. Azure is running on its technology based on Hyper-V.
We have disabled SELinux for all the servers, install FIO and reboot the servers to make sure SELinux is disabled:
yum -y update sed -i -e 's/SELINUX=enforcing/SELINUX=disabled/g' /etc/selinux/config yum -y install fio git gcc screen
[reboot]
Run the two tests:
- Run Unix Bench:
cd; mkdir unixbench; cd unixbench
git init; git clone https://github.com/kdlucas/byte-unixbench
cd byte-unixbench/UnixBench
screen -S benchmark
./Run
2. Run fio stats:
# Random read/write performance test
fio --randrepeat=1 --ioengine=libaio --direct=1 --gtod_reduce=0 --name=test --filename=random.fio --bs=4k --iodepth=64 --size=4G --readwrite=randrw --rwmixread=75 | grep IOPS; rm -f random.fio
fio --randrepeat=1 --ioengine=libaio --direct=1 --gtod_reduce=0 --name=test --filename=random.fio --bs=256k --iodepth=64 --size=4G --readwrite=randrw --rwmixread=75 | grep IOPS; rm -f random.fio
fio --randrepeat=1 --ioengine=libaio --direct=1 --gtod_reduce=0 --name=test --filename=random.fio --bs=4m --iodepth=64 --size=4G --readwrite=randrw --rwmixread=75 | grep IOPS; rm -f random.fio
# Random read performance test
fio --name=randread --ioengine=libaio --iodepth=16 --rw=randread --bs=4k --direct=1 --size=4G --numjobs=2 --runtime=240 --filename=random.fio --group_reporting | grep IOPS; rm -f random.fio
fio --name=randread --ioengine=libaio --iodepth=16 --rw=randread --bs=256k --direct=1 --size=4G --numjobs=2 --runtime=240 --filename=random.fio --group_reporting | grep IOPS; rm -f random.fio
fio --name=randread --ioengine=libaio --iodepth=16 --rw=randread --bs=4m --direct=1 --size=4G --numjobs=2 --runtime=240 --filename=random.fio --group_reporting | grep IOPS; rm -f random.fio
# Random write performance test
fio --name=randwrite --ioengine=libaio --iodepth=1 --rw=randwrite --bs=4k --direct=1 --size=4G --numjobs=2 --runtime=240 --filename=random.fio --group_reporting | grep IOPS; rm -f random.fio
fio --name=randwrite --ioengine=libaio --iodepth=1 --rw=randwrite --bs=256k --direct=1 --size=4G --numjobs=2 --runtime=240 --filename=random.fio --group_reporting | grep IOPS; rm -f random.fio
fio --name=randwrite --ioengine=libaio --iodepth=1 --rw=randwrite --bs=4m --direct=1 --size=4G --numjobs=2 --runtime=240 --filename=random.fio --group_reporting | grep IOPS; rm -f random.fio
The Unbelievable Result
Seeing is believing so here is the comparison of the two tests between the tested machines.
IOPS result says that both AWS and Azure are particularly conservative (or shall we say stingy?) on IOPS which will result in lower performance.
This also explains the Unix Bench results where the overall performance of the AEserver Cloud beats the Hyperscalers clearly
To conclude, hype and trendy doesn't always mean the best. :)
Now let's look at the price comparison of these machines (Currency is UAE dirhams at a rate of 1USD=3.67AED)
Notice: Bandwidth is not included for both AWS & Azure! Adding 5mbps bandwidth -take 2TB per month of transfer would result in an additional of at least AED 365/$100 (considering a very conservative rate of US$0.10/GB).
Final Verdict: AEserver's UAE Public Cloud is not only 100% local Dubai based, but beats the big guys in performance and IOPS at a whopping 50% less cost!
Are you are looking for that sweet spot of high performance and value? Wanna make sure that your investments are the best ones? Go check our CLOUD SERVERS now!
Let us know that you have read this article. Order now a cloud server using the promo code HYPE5 for a 5% discount.
