Chaos Engineering in the Cloud with Keysights' Cyperf Solution☁️
Chaos Engineering is a niche-focused on the simulation of distributed services on the cloud, with an end goal to find weaknesses in the system before they cause problems for your customers, hence improving the system availability.
During Tech Field Day 25, Keysight, an internet measurement company presented its products to we the delegates.
Keysight Solution Overview
Some challenges Enterprises suffer when building cloud solutions are:
Scale: Most tech startups are going to Kubernetes first. Leveraging Keysights agent sizing that’s completely elastic, services can be simulated to scale.
Going Multi-Cloud. I admire how Keysight is supported on Single Cloud, Multi-Cloud, and Hybrid Cloud. Even on-premises (During the presentation on-premises was used)
As Microservices call other services, issues like Network Latency, Congestion, or failure to reach the other service can occur. Using a test solution to simulate these issues is pretty neat.
Failure in one microservices which then affects other connected services can cause chaos. It’ll be great to simulate this type of use case beforehand.
Client PCs with apps like Skype running on them can be simulated too so the traffics impact on the network can be observed.
In the above image which was taken from the Keysights presentation, we can see a simple UI on the far left describing Apps that can be simulated.
It’s pretty neat how this is possible, specifically on the cloud. Why am I saying this? A few months ago while researching Cloud measurements for understanding the performance of workloads running on the cloud. I stumbled upon a presentation which said the cloud isn’t a great place to do measurements.
It’s interesting how Keysight made this easier with their solution. They made it possible to use their solution to simulate workload network traffic, which leads to collecting performance information of the network.
I did a little lab on my AWS account where I deployed the Keysight solution and it was successful.
In case you want to get a feel of it on AWS simply follow these simple steps:
Download the Cyperf Cloudformation JSON script
Follow the simple steps on the README
Back to the Cyperf’s presentation on Techfield Day where the client app traffic was simulated.
Below we can see at the centre of the image, that simulated client traffic (7.11GB) was generated.
Conclusion
I love the solution so much and I very much see a few enterprises adopting it soon, it saves the Cloud engineer time because failovers will be less. If I can simulate apps running on my cloud infrastructure ahead of time, I can write automation scripts to keep the infrastructure up when there is a failover caused by these apps.
Check out the full product demo! <3