COVID-19: Kubernetes Helps Retail, Gaming Meet Sudden Scaling Demands
As people around the world have adapted to the realities of stay-at-home orders and quarantines, nearly everyone has found themselves relying on technology more than ever. In some cases, that means traffic to websites or applications scaling 10 times or more nearly overnight. It should be “In other cases, companies that have seen dramatic decreases in foot traffic have turned to technology as a way to maintain their connection to customers.”
We chatted with Michael Ferranti, vice president of marketing at container-native storage company Portworx, which offers a Kubernetes storage platform, about the trends Portworx has seen customers grapple with as they both scale rapidly and launch new apps to meet novel demands.
How have Portworx customers been effected by the COVID-19 pandemic? What trends have you seen?
First of all, a lot of our customers are in “wait-and-see” mode to see what the overall impact on the economy will be. But there are three types of customers who are seeing huge, rapid scaling. They are retail companies, gaming companies and data analytics.
We have two retail customers who have pretty interesting use cases. One is a national grocery retailer that operates a number of regional chains, and they built an online inventory search function for each brand’s website. Basically, it will let you go to the website and see if they have hand sanitizer and toilet paper in stock right now. They had already rolled out this feature, and it was already designed to be scalable — traffic is very different right before Thanksgiving as it is on October 15th. This feature is built using Kubernetes and Elasticsearch, and they use Portworx for the underlying storage platform.
The nice thing about Kubernetes is that it’s basically a scale out model, so they don’t have to re-architect the application in order to hit that demand, they just need to add more servers.
Once the COVID-19 outbreak started, they started seeing 10 times the number of queries per hour. I think they’re scrambling a bit, but it’s not because things aren’t working. The nice thing about Kubernetes is that it’s basically a scale-out model, so they don’t have to re-architect the application in order to hit that demand, they just need to add more servers. There are certainly some operational implications, but it’s much less than if you needed a new architecture to meet the new demand.
The other retail customer is in Europe, which as we know has been pretty hard hit. So this customer runs both traditional grocery stores as well as a network of smaller markets that might be attached to a filling station. As the situation in Europe became more serious, this retailer decided to partner with an Instacart-like service to deliver groceries, but from the smaller markets, that were often located outside of big metro areas, as a way to get food and other essentials to distributed populations.
I believe that this app is either in production now or going into production really soon. Even if we want to be really generous and say they started on March 1, in enterprise software development it’s really breakneck speed to go from idea to large-scale rollout in 30 or even 60 days.
Other than retail, what other interesting examples of scaling have you seen?
We’ve all been watching all the models unfold and tracking which states have the most cases and the most deaths. We have a client that provides geo-analytics and data analytics and they are seeing a lot of growth right now. The models are being essentially updated in real-time as data comes out of the [U.S. Centers for Disease Control] and used to give policymakers and healthcare professionals real-time data about the COVID outbreak.
We also have a client in the online gaming industry, and it saw a five or six X increase in platform usage pretty much overnight. They’ve been able to manage scaling relatively seamlessly using what they call points of presence. So if they have a bunch of new people playing in Milwaukee, they’re going to go into a colocation facility in Milwaukee and set up a containerized cluster in that facility that has all the game services available. If usage goes down, they remove that point of presence. So the architecture really lends itself to spiky usage.
Speaking of architectures, how have the architectures these companies use helped them scale?
Some people wonder if there might be too much hype around Kubernetes, but I think that Kubernetes is really shining right now. When Kubernetes is applied to the type of problems it was designed to address, like massive scaling, it performs very well. These retailers and gaming platforms were already designing their applications to accommodate spikes in usage, just not for this kind of spike. Just the fact that with Kubernetes, you can handle ten times as much volume pretty much overnight just by adding some more servers and scaling out your Kubernetes cluster, I think that is a huge endorsement for Kubernetes when applied to the right types of projects.
Portworx is a sponsor of The New Stack.
Feature image by FunkyFocus from Pixabay.