Automating Node Provisioning: How The Analytics Giant NielsenIQ Saved Up To 80% On Cloud Costs
~44, 000 employees
New York City, USA
Cloud services used
Azure Kubernetes Service (AKS)
Delivering intelligence in the cloud
NielsenIQ is a global leader in data analytics and audience insights that empower companies with actionable intelligence and help them connect with audiences. The organization operates in more than 55 countries and employs c. 44,000 people.
To deliver accurate and timely intelligence, NielsenIQ relies on a number of cutting-edge technologies. Among them is the growing adoption of Kubernetes. However, NielsenIQ teams were lacking process control measures in certain areas, which translated into growing cloud costs.
“That’s our trajectory now: your team migrated to Kubernetes, so now it’s time to follow our best practices. To address the cost issue and provide our teams with the right tooling, we started looking into various solutions created specifically for Kubernetes,” said James O’Hare, Principal Platform Engineer at NielsenIQ.
The team considered the open-source node provisioning tool Karpenter. But it quickly turned out that the solution wouldn’t work in the company’s cloud ecosystem based on Microsoft Azure services. After assessing other cost-related Kubernetes offerings like Kubecost, the team found CAST AI.
We didn’t come to CAST AI for cost at all, but for orchestration of node pools. This was our most pressing issue at the time. Cost-wise, we previously looked at Kubecost, but CAST AI has an overall application that has really helped us in a bunch of areas. It allowed us to be very agnostic in regards to how we have our cluster structure set up and stamp clusters following a simple pattern.
Normally, we’d need to take time to find a good fit between cloud resources and workloads. With CAST AI, we can avoid a lot of that and adjust the capacity based on the analysis of workload requirements. This is where we saw the value of CAST AI. And then we saw that it’s generating significant savings.James O’Hare, Principal Platform Engineer at NielsenIQ
Implementing CAST AI was a breeze
The NielsenIQ team reached out to the CAST AI team and started the implementation.
We ended up talking to one of the co-founders of the company, and that was pretty nice. The VP of Customer Success helped us get much stuff done. We were having some sort of production issue early on, and people were instantly available on Slack, ready to jump on a call and start solving the problem right away. So, we had a good experience with the support, and since the implementation, things have been running smoothly.James O’Hare, Principal Platform Engineer at NielsenIQ
“I think it’s been a good relationship. It has been great to work with the CAST AI team, especially the founders. I totally get where they’re coming from, and building CAST AI was a great idea. I wish them all the success because it’s such a great tool that has proven to be very helpful for us,” added James.
Achieving 80% of cost savings and realizing ROI after two months
By implementing CAST AI, NielsenIQ generated 60–80% cost savings on their non-production deployments and 40–50% savings for production clusters.
Now that we’re seeing such large savings on the non-production side of the fence, we’re forcing a lot of teams to leverage spot VMs more. CAST AI handles all the provisioning automatically. So, we’re saving tons of money, and by leveraging CAST AI’s automation features we’ve been able to move many teams to spot VMs.James O’Hare, Principal Platform Engineer at NielsenIQ
NielsenIQ saw benefits from implementing CAST AI very quickly and their investment was returned within two months.
CAST AI paid for itself close to within the first two months thanks to all the savings it generated around compute costs.James O’Hare, Principal Platform Engineer at NielsenIQ
Get results like NielsenIQ – book a demo with CAST AI now
Automated node provisioning was a game changer for NielsenIQ
As every other major cloud provider, Azure offers a large number of instance types and sizes. Considering the sheer scale of Azure’s offering, picking the right machine for a Kubernetes application is a challenge for many teams.
By implementing CAST AI, NielsenIQ teams don’t have to worry about choosing the best machine to match the cost and performance objectives.
This feature of CAST AI has been a lifesaver for us. It enabled us to create an architectural pattern that has made it very easy to stamp and continue rolling stuff out. There’s a lot of extra features that come with that, but for us, it’s the node provisioning automation that forms the heart and soul of the application.James O’Hare, Principal Platform Engineer at NielsenIQ
The feature had a positive impact on the engineers’ workload and well-being. “CAST AI enabled us to standardize our workloads and automate the entire process, which had a big impact on the amount of work facing our engineers,” said James O’Hare.
CAST AI seamlessly fits Nielsen’s cloud cost management process and toolkit
The team at NielsenIQ supports other teams as they move to Kubernetes and embrace the toolkit that includes CAST AI for better cost control and visibility.
We’re a team addressing the Kubernetes footprint for the entire global enterprise. We leverage CAST AI to be able to put that footprint in place, especially as we’re seeing more and more teams looking to migrate to Kubernetes. One of the things CAST AI enabled us to do is carry out a cost analysis for a team and show the estimated cost savings to prove the value they would get from following our practices and tooling.
It allows teams to come in and actually see what they could save – I just tell them to go ahead, connect their cluster to CAST AI, and have the read-only agent scan their cluster to find out potential cost savings.James O’Hare, Principal Platform Engineer at NielsenIQ
NielsenIQ is planning to advance its migration to Kubernetes and CAST AI will be an integral part of the tooling ecosystem offered to teams. “As we roll out more and more clusters, we will achieve greater savings. We are offering CAST AI as part of our best practices and tooling package for teams joining our enterprise,” said James.
What CAST AI allows us to do is give the app development teams the information that they need they can show to product leadership and say, ‘Hey, we can save a lot of money if we go down this route, so let’s move to Kubernetes.’ Our FinOps specialist drives much of that effort and uses CAST AI to get the insights he needs.James O’Hare, Principal Platform Engineer at NielsenIQ
Get results like NielsenIQ – book a demo with CAST AI now
CAST AI features used
- Spot instance automation
- Real-time autoscaling
- Instant Rebalancing
- Full cost visibility