It can seem like a no-brainer to migrate cloud computing environments over to the next generation of hardware and performance, like upgrading the C3 to C4. Let’s take a look at possible conditions that might make a team reconsider staying with the previous generation.
Looking to compare the newer C4 and C5 instances? Check out the Amazon EC2 Comparisons: C4 vs. C5 post!
It can seem like a no-brainer to migrate cloud computing environments over to the next generation of hardware and performance. One case could be considering the switch from AWS EC2 c3 to c4. But, let’s take a look at possible conditions that might make a team reconsider staying with the previous generation.
We’ll start by taking a look at the specs and difference between the two generations. And then, we’ll focus on storage strengths and weaknesses as a key way to identify if the upgrade is a right move.
A look at specs and pricing
The c4 lineup features identical memory and available CPU cores as the previous c3 generation. From the table below, available compute units, network, and EBS throughput stand out as the most visible difference-makers.
|Instance||Memory||Compute units||Cores||Storage||Network Performance||EBS-optimization|
|c4.large||3.75 GB||8 units||2||EBS only||Moderate||500 mbps|
|c4.xlarge||7.5 GB||16 units||4||EBS only||High||750 mbps|
|c4.2xlarge||15 GB||31 units||8||EBS only||High||1000 mbps|
|c4.4xlarge||30 GB||62 units||16||EBS only||High||2000 mbps|
|c4.8xlarge||60 GB||132 units||36||EBS only||10 Gigabit||4000 mbps|
|c3.large||3.75 GB||7 units||2||32 GB (2 * 16 GB SSD)||Moderate||n/a|
|c3.xlarge||7.5 GB||14 units||4||80 GB (2 * 40 GB SSD)||Moderate||500 mbps|
|c3.2xlarge||15 GB||28 units||8||160 GB (2 * 80 GB SSD)||High||1000 mbps|
|c3.4xlarge||30 GB||55 units||16||320 GB (2 * 160 GB SSD)||High||2000 mbps|
|c3.8xlarge||60 GB||108 units||36||640 GB (2 * 320 GB SSD)||10 Gigabit||n/a|
|Instance||Linux On-Demand||Linux Reserved Instance||Windows On-Demand||Windows Reserved Instance|
Pricing wise, the c4 packs more punch per dollar spent, especially after a recent price reduction during early 2016, while featuring an overall spec increase from processor type, allocated ECU, and EBS optimizations. What’s interesting to note for Linux users is that the on-demand rates are slightly cheaper for c4 instances (except the large), but have slightly more expensive Reserved Instance rates when compared to c3 RI prices.
[Edit: 11/14/2016] AWS will make another C4 pricing adjustment in 2016. As of December 1, 2016, there will be various C4 price reductions (anywhere from 5 to 20 percent) within certain regions.
As always, Windows OS users have higher hourly operations costs. But, overall, it’s cheaper to run the c3 family. Doing so means missing out on the upgraded hardware and increased EBS throughput of c4.
The c4 family provides the same 64-bit architecture, but with a high-frequency Intel Xeon E5-2666 v3 (Haswell) processor. AWS advertises these as being optimized specifically for EC2 use. Meanwhile, the c3 lineup features the previous-generation, high-frequency Intel Xeon E5-2680 v2 (Ivy Bridge) processors.
Not only is the processing hardware different, but the c4 lineup offers the ability to alter the C-state and P-states of the cores. The ability to program the idle and busy states of the processors can yield big operational improvements.
Where the c4 lineup starts to differentiate is in higher EBS throughput within the family, at lower prices across the board. Using c4 also requires a VPC-only setup.
Consider this: specific storage needs can determine whether to upgrade
Unlike its newer sibling, the previous-gen c3 features ephemeral storage using local SSDs. Specs-wise, the local SSD storage provides faster speeds than EBS’ provisioned throughput rate, as it’s attached to the instance and doesn’t require network bandwidth to send and receive data.
If workloads require a super high rate of read or write operations to its volumes then the c3s may be the only choice when comparing these two families. We would recommend running a side-by-side test of comparable c3 and c4 instances for further performance metrics. However, when the instances shut off, all data on the c3’s local storage disappear, as there is no persistence of data with the onboard SSDs.
Is EBS optimization required?
The c4 lineup features EBS optimization at no additional charge, unlike the c3 lineup which has various charges per hour (see table below). If users require EBS optimization for their environments, this will have the biggest impact on price. It will make the c4 the more cost-effective option.
EBS pricing examples for c3
|Instance||EBS pricing per hour|
Working with EBS volumes means data always persists on the attached volume. This opens the cost savings opportunity of programmatically scheduling c4 instances to shut off and on to optimize EC2 utilization. This is doable with the c3 family as well if there’s an attached EBS volume, but, with the caution of losing data attached to the local SSDs during shut off.
Which is best from a cloud cost management POV?
According to our technical account managers, it comes down to this: if users need the hardcore IOPS of local storage, the c3 will be the only practical choice. If users require EBS optimization, c4 will be the best bet as it will be cheaper and gives higher EBS throughput.
If users don’t require EBS optimization, they can go either way, but the extra compute of the c4s can be a persuading factor. However, if extremely high disk I/O to local SSD storage is a must, it might be more practical to stay with the c3 instances. This includes the need to process high volumes of log files and other user streams of data. It’s where the c3 can still outshine the c4 for now.
For everyone still considering between the two families, visualizing EC2 and storage cloud costs and usage can shed a lot of light and insight on the debate. We welcome anyone interested in building a better look at their cloud cost infrastructure to reach out to us for a trial. The right cloud cost reporting and visualizations can show AWS cost and usage efficiency like never before, allowing IT, operations, and finance professionals to make sound decisions when growing and optimizing their AWS environments.