Amazon EC2 Comparisons: C5 Instances

C5 instances are the current generation of compute-optimized EC2 instances. When compared to the general-purpose M5 instances, C5 has more compute power, a lower memory/vCPU ratio and a lower price point. For example, the m5.xlarge has four vCPU and 16 GiB of memory for an On-Demand price of $0.192/hr, while the c5.xlarge has four vCPU and eight GiB of memory, but a lower price of $0.17/hr. As a note, these prices are for Linux instances in US West (Oregon). (Learn more in our C5 vs. M5 blog.)

For compute-intensive workloads, C5 is generally going to be the better choice, but there are a few different options. Let’s take a look at C5, C5d and C5n to see how they compare in specs and cost.

Amazon EC2 C5 Specs

The first C5 instances were released in November of 2017. The instances are built on 3.0 GHz Intel Xeon Scalable (Skylake) processors, and have the potential to run at speeds up to 3.5 Ghz using Intel Turbo Boost Technology. C5 was also the first generation of compute-optimized instances to incorporate the AWS Nitro system, which includes the Nitro hypervisor, Nitro card and Nitro security chip.

C5 instance specs. Prices are for Linux instances in US West (Oregon).

C5 was well received, and within the first year of its release AWS released two new compute-optimized instances — the C5d and C5n.

Amazon EC2 C5d Specs

Released in May of 2018, the C5d has the same basic structure as the C5. The vCPU, memory and network specs are all the same, but the C5d adds on local NVMe-based SSD block level storage. Since it’s physically connected to the server, the local storage has higher transfer speeds and lower latency than EBS storage. The SSD storage is also tied to the instance, which keeps unused volumes or snapshots from sticking around once the instance is spun down. This can be a hazard with EBS instances, since EBS volumes are persistent beyond the lifespan of the instance. (Find out more about controlling EBS costs.)

C5d instance specs. Prices are for Linux instances in US West (Oregon).

The attached storage makes C5d a good choice for media processing, such as video encoding or image manipulation. Other good use cases include applications that require high levels of cache or need temporary storage of data, like batch processing.

Amazon EC2 C5n Specs

Released at AWS re:Invent 2018, C5n instances are made for processing massive amounts of data. Built with the same processor as other C5 instances, the C5n family incorporates more memory. Unlike the C5 or C5d, which can use up to eight queues for packet processing, the C5n Elastic Network Interface (ENI) can make use of up to 32 queues, giving C5n instances a high amount of network bandwidth — up to 100 Gbps on the c5.18xlarge.

C5n instance specs. Prices are for Linux instances in US West (Oregon).

The memory and network bandwidth combination allow C5n instances to run compute workloads with massive amounts of data, such as machine learning, tapping into data lakes, analytics or High Performance Computing (HPC).

Comparing AWS EC2 C5 costs

With any EC2 instance, the goal to cloud cost optimization and increased efficiency is to find the perfect balance between underprovisioning and overprovisioning. The three C5 options make it easier to find that balance with your compute workloads.

First, let’s see how they stack up directly on costs:

Prices are for Linux instances in US West (Oregon).

C5 has the cheapest rate, but that doesn’t mean it’s always the most affordable option. It really comes down to the workload and compute task. Say your data is going to be pulling from S3, run through the compute instance, then deposited back in S3. Running C5d without an EBS volume might be your best bet. Not only would you save the cost of the EBS volume, but the higher transfer speeds decrease your compute time, allowing you to turn off the instance sooner.

At the same time, C5n might be a better bet, especially if you’re pulling large amounts of data out of S3. Sure, C5n costs 27% more than C5 to run per hour, but a network bandwidth that’s 2.5-4x more with increased memory translates to faster compute times. With a heavy compute load, all you have to do is cut the compute time by 27% or more to start saving money.

In the end, you should incorporate a tool into your process that allows you to make these calls based on your workloads and your compute use, such as Cloudability’s Rightsizing feature. This feature weighs your workloads against the capability of your EC2 options, cross-references them with price and gives you recommendations based on clipping risk.

Ready to start optimizing your cloud costs by finding the exact right EC2 instance? Sign up for a free trial today!