In this double-interview, HostingJournalist talks with Liz Rice, the Chief Open Source Officer at Isovalent, and Donnie Berkholz, the Senior Vice President of Engineering at Percona. We’ll discuss the future of platform engineering and its impact on how companies run their data centers and hosting arrangements.
In this conversation, we talk about how platform engineering can help businesses cut expenses, boost productivity, and better the developer experience. Liz Rice and Donnie Berkholz also tell us of the value of open-source components in preventing vendor lock-in and the part Kubernetes plays in allowing uniformity and automation across various server settings. In this interview, they stress how important it is for platform tech teams to adopt cloud-native principles and implement more automated, agile, and secure security practices in conventional infrastructure.
Can you tell us what platform engineering’s future holds?
Liz Rice, Isovalent: Platform engineering is about creating a reliable, consistent platform so developers can focus on their work. eBPF can help enable this by enabling instrumentation at the platform level, reducing the need for application-specific code changes for things like observability and security. We’ll see platform engineering teams increasingly bringing cloud native principles to traditional infrastructure, with increased agility, automation and enhanced security practices.
Donnie Berkholz, Percona: In 2023, more companies will care about controlling their cloud spend so they can operate efficiently. Making use of a platform engineering approach should help them keep that positive developer experience and efficiency that cloud offers, but also help them to avoid or reduce costs and lock-in.
Will platform engineering approaches affect how companies run their data centers and hosting arrangements?
Donnie Berkholz: Platform engineering helps developer teams manage their experience and improve efficiency, based on the lessons learned across DevOps, Site Reliability Engineering and cloud deployment. It covers how teams can deliver the right kind of developer experience using automation and self-service, so developers can get to writing code and implementing applications rather than having to wait for infrastructure to be set up based on a ticket request.
Companies will adopt this approach because they care about their internal developer experience – anything that gets in the way of those developers is literally costing you money when those employees are not productive. When your developers can move more quickly, they can iterate faster and deliver better applications or services to you. When companies care more about their capital expenses, any approaches that can help them reduce cost from licences and free up time will be welcomed quickly.
Liz Rice: Absolutely. We are already seeing platform teams that started out building cloud-only infrastructure, and are now expanding back into on-premise enterprise infrastructure. Platform engineering needs to make this transition as seamless as possible for their development teams.
How does the growth of Kubernetes affect how and where companies decide to run their operations?
Donnie Berkholz: Kubernetes fits alongside platform engineering because it provides that standardised approach to run and orchestrate your systems. While it started out aimed at stateless applications, it now can be applied across stateful services like databases that you will see running for long periods of time continuously. Developers want to run and manage those applications and infrastructure components together, and Kubernetes is where they intend to do that.
Alongside this you have the trend towards more microservices architectures – developers want to build smaller components and connect them up with APIs, and running containers with Kubernetes makes the management side easier to automate.
Liz Rice: Kubernetes has become the defacto standard, making it easier to run (almost) the same infrastructure in clouds and on premise. Data gravity, regulations, customer requirements, and other factors will determine exactly where workloads run, but the growth of Kubernetes and cloud native tools like Cilium makes it easier to run consistent platform tooling everywhere.
Donnie Berkholz: What does this mean for infrastructure? It should make it easier to decouple your applications from where you run them, as long as you use open source components. If you use specific cloud service providers’ offerings, then you will be tied to that provider. There’s no right or wrong answer here – you might want to take advantage of what Microsoft, Google and AWS provide because it is the right design for you. But that doesn’t mean that everything should be tied to those providers – running standard workloads like databases that you can run where you want to can help you reduce costs overall, compared to spending on specific database as a service offerings that claim to be open-source compatible when they actually are not.
What challenges do companies face around their data center planning, and how can platform engineering improve the situation?
Liz Rice: Platform engineering teams from the cloud native world have completely different expectations about the scale of deployments they can manage, and the automation required to do so. Platform teams of half a dozen in strength are used to running cloud infrastructure with thousands of virtual machines. As cloud-first tools like Cilium make it easier to integrate workloads on Kubernetes with legacy workloads, this reduces complexity, enabling platform engineering teams to bring increased automation and improved operational efficiency into their networks and data centers.
Donnie Berkholz: Supporting platform engineering makes it easier for your team to be efficient while also keeping the benefits of open source around running things how you want to, rather than how a provider prescribes you should. It helps companies that want to see the benefits of all the great steps taking place around cloud, but that can’t or won’t run in the cloud. For companies weighing up their options around cloud, hosting or running their own environments, platform engineering keeps those options open in all senses of the word.
About Percona and Isovalent
Percona is a provider of open-source database software and services. The company was founded in 2006 and is headquartered in Raleigh, North Carolina, with offices and employees around the world.
Percona offers a variety of solutions for managing, optimizing, and scaling open-source databases such as MySQL, MongoDB, and PostgreSQL. These solutions include software products such as Percona Server for MySQL, Percona XtraDB Cluster, and Percona Monitoring and Management, as well as professional services such as consulting, training, and support.
Isovalent is a software company that specializes in developing and providing enterprise-grade networking solutions for Kubernetes and cloud-native environments. The company was founded in 2018 and is based in Berlin, Germany, with offices and employees in the United States as well.
Isovalent’s flagship product is called Cilium, which is an open-source networking and security solution for Kubernetes. Cilium provides a “highly scalable and performant” network security model by leveraging eBPF (extended Berkeley Packet Filter) technology, which enables fine-grained filtering and policy enforcement for network traffic.
