Shopping experience in the digital age

Voggt is a livestream shopping app for pop culture fans in Europe. The company provides a platform that connects collectible buyers and sellers through live video auctions, attracting hundreds of thousands of users across the continent. 

The unique nature of Voggt’s product places specific requirements on its underlying infrastructure.

An auction must happen in near real time – when you bid on something, you want to do it now and not one second later. It’s very important to have a very low latency between our users and the backends, where possible. So we need to have an infrastructure that is high-performing, efficient, and very fast.

Nicolas Hug, Lead SRE at Voggt

Running multiple Kubernetes clusters became challenging

Kubernetes is a go-to solution for startups like Voggt that are looking to create flexible and scalable applications. However, managing clusters, nodes, and pods quickly became a burden as the company’s Kubernetes footprint grew. 

I was using Kubernetes to schedule workloads, but the process wasn’t fast enough. Sometimes, it took almost two minutes to get a new machine in the cluster. This just wasn’t acceptable when we had a lot of traffic coming to the platform. Now with CAST AI, it takes less than 1 minute.

Nicolas Hug, Lead SRE at Voggt

This prompted a search for an automation solution 

The company started searching for an automation solution that would help solve the scheduling problem. At first, they tested Spot by NetApp. 

I found CAST AI and immediately thought it looked more approachable. I wasn’t wrong, as the onboarding was very fast with CAST AI. We were running it in production within one week, while with Spot by NetApp, it was endless testing and support calls.

Nicolas Hug, Lead SRE at Voggt

Going from zero to CAST AI in production in a week 

Voggt could easily procure CAST AI thanks to the solution’s availability on the Google Cloud Marketplace. “We purchased CAST AI via Google Marketplace, which made the procurement process much easier for the billing team. That way, they can centralize all the CAST AI bills on the Google Cloud Platform. It was very easy to make it all work,” said Nicolas Hug.

The company carried out a full implementation of CAST AI within a week, running it across all of its environments.

I like to test things fast. When I met one of the CAST AI co-founders, I instantly trusted him. I trusted CAST AI, so I thought: Let’s try it, and if it fails, it’s not very important. We’re a startup and can allow it to fail right now. But it didn’t – CAST AI worked really great from day one.

Nicolas Hug, Lead SRE at Voggt

The CAST AI team supported Voggt at every step of the way.

The support we received was marvelous. I think it was one of the best onboarding experiences and the best support experience I ever got. The CAST AI team was very reactive, which is especially important when starting a project because we need to move fast and can’t wait days to get our question answered. The team was very nice, and it was a pleasure to work with them during our onboarding.

Nicolas Hug, Lead SRE at Voggt

Reducing cloud costs by 25% with spot instance automation

It took Voggt just one week to implement CAST AI, and the team has seen immediate ROI. 

CAST AI paid for itself almost by the first day it was activated in production.

Nicolas Hug, Lead SRE at Voggt

CAST AI opened the door to Spot VMs, the cost-effective compute instance type that may get interrupted at any moment. “Before CAST AI, we only used compute at a full price. CAST AI allowed us to use Spot VMs, which can shut down at any moment – but with CAST AI, this is all fully managed. And if there’s no more compute available, the platform will smartly use other machine types,” said Nicolas Hug.

We got some very good cost improvements because CAST AI did a very good job compacting our Kubernetes cluster and making sure that we’re using only what we need to use. We reduced our cloud costs by 25% using CAST AI, which is great and very important for us since we’re a startup and costs are very important for investors.

Nicolas Hug, Lead SRE at Voggt

Massive time savings thanks to automated cluster scheduling

CAST AI also solved the core infrastructure pain Voggt experienced: the sheer amount of manual work it took to schedule workloads and manage nodes. Implementing it helped the company’s Lead Site Realiability Engineer to entirely automate this process.

After implementing CAST AI, I finally had the time to focus on developer experience at Voggt or to think about the future of our infrastructure, for example, when we decide to target markets outside of Europe. Or simply the things we could do to have a better infrastructure and deliver a better product for our end users.

Nicolas Hug, Lead SRE at Voggt

He added: “Before CAST AI, I had to manage nodes myself. Now, it’s not my problem anymore. I can do other things that are more important than workload scheduling and thinking about how many nodes I want, what size they should be, etc. I fully trust CAST AI to make the best choices here. It’s a massive gain in time for me.”

Next steps

As Voggt grows its platform and user base, CAST AI will be there to support the company’s expanding Kubernetes footprint, enabling automation across the board. 

Everybody – from a small startup to a large company – should use a product like CAST because it comes with too many benefits to miss.

Nicolas Hug, Lead SRE at Voggt