Amazon Web Services ECU vs vCPU — All You Need to Know
You have undoubtedly used an EC2 instance if you have deployed an application or service to the Amazon Web Service (AWS) cloud. Which instance type to use was one of the considerations you had to make before you could start a new instance. Selecting an EC2 instance type might be a difficult task. AWS divides its instance types into instance families, ranging from micro to 32xlarge within each instance family.
The availability of instance types varies by region. However, the primary factor that distinguishes each instance type is a combination of the following:
· Compute power or CPU
· Memory
· Storage capacity and type of storage
· Network connectivity and performance
We are going to look at compute power in this essay. We will discuss the difficulties of measuring and comparing compute power in a virtual environment and the various ways AWS has attempted to convenient for users to make intelligent decisions about instance types.
The Challenge Presented by the Virtual Environment
When you create an EC2 instance, you are renting space in one of AWS’ data centers. Virtual machines called EC2 instances are hosted on powerful servers that you share with other AWS users. One of the advantages of the Virtual machine model is that you, as a consumer, need not be aware of the underlying server architecture specs.
The problem with this is that comparing instance types based on their specifications might be difficult. You get a share of the processing power from the host machine when you use a virtual machine. AWS needed the means to measure this allotted capacity and compare different instance types because an allotment of processing power is not the same as a real processor. Compute power was provided as an ECU value when EC2 first released (or an EC2 Compute Unit).
What is an EC2 Compute Unit?
The idea for an ECU emerged from Amazon’s need to guarantee continuous CPU capacity for their instances, regardless of the underlying hardware. A single EC2 compute unit has the same processing power as a 2007 Intel Xeon or AMD Opteron processor running at 1.0–1.2 GHz.
The AWS ECU was the standard measurement of processing power on all EC2 instances from 2006 until 2014. This, for the most part, addresses the issues with comparing different instance types, allowing AWS to set instances on newer infrastructure while still providing the same capacity as instances on older hardware.
Benchmarking at the time indicated that ECU measurement was generally equal. However, they did observe a performance boost in direct correlation with the number of ECUs. According to one theory, this improvement could be attributed to a reduction in the number of instances hosted concurrently on a host server. The results of one of these benchmark studies can be seen here.
ECU or vCPU?
One of AWS’s issues was educating new and existing customers about the ECU metric, which was not always a straightforward idea for specific customers to grasp. Virtual machines have traditionally been quantified in virtual CPUs (or vCPUs); hence AWS replaced the ECU metric with the more conventional vCPU in 2014.
The ECU established a consistent approach of measuring processing performance across all instance types, which was one of the advantages of utilizing it to assess processing power. However, because it is not a standardized term, it has a high risk of being misinterpreted.
Factors Which Can Affect The Power of Your vCPUs
The most critical element affecting an instance’s processing power is how the vCPUs are configured on the instance type. Let us compare the compute-optimized compute-optimized C4 instance types to the general-purpose M4 instance types. You will note that the processor specs are not the same.
C4 Compute-Optimized — 2.9 GHz Intel Xeon E5–2666 v3 Processor — Intel AVX†, Intel AVX2†, Intel Turbo — EBS-optimized — Enhanced Networking
M4 General-Purpose — Up to 3.1 GHz Intel Xeon Platinum Processor — Intel AVX†, Intel AVX2†, Intel Turbo — EBS-optimized — Enhanced Networking
The underlying hardware is another aspect that can affect processing performance. You may find out the precise CPU running on the host computer by running the following command on a Linux instance.
$ cat /proc/cpuinfo | grep “model name”
If you run this command on two different machines, you might get something like this:
i-5560…. model name: Intel(R) Xeon(R) CPU E5–2650 0 @ 2.00GHz
or
i-bb23…. model name: Intel(R) Xeon(R) CPU E5506 @ 2.13GHz
If performance is necessary one approach runs a similar command when creating a new machine, rejecting instances created on older hardware.
Summary
It is critical to grasp the significance of the offered measures for your instance type, whether you are talking about ECUs or vCPUs, and how they compare to physical processor kinds. If performance is critical, you will want to read over the precise specifications for each instance type to ensure you are getting the most excellent processing power for your application.
