Skip to main content

4 posts tagged with "cloud"

View All Tags

Lars Kamp

Dvir Mizrahi is Head of Financial Engineering at Wix, the leader in website creation with 220 million users running e-commerce operations. And with over six thousand employees, Wix ships more than fifty thousand builds each day.

Dvir is also among the original authors of the AWS Cloud Financial Management certification.

In this episode, Dvir covers how Wix shifted from FinOps to Financial Engineering. It's an engineering-first approach to build tooling and processes tracking financial key performance indicators (KPIs) for its multi-cloud infrastructure. The new approach established a culture of financial responsibility that supports Wix's continued growth.

Wix started in 2006 and initially ran its infrastructure on-premise. Today, Wix runs a multi-cloud environment on Google Cloud Platform (GCP) and Amazon Web Services (AWS). As Wix shifted from on-premise to the cloud, the procurement process of resources changed with it.

In the old world, purchasing additional hardware was a closed and controlled process. But in the cloud, Dvir compares resource procurement to "a supermarket where people can go in, take whatever they want, and leave without passing the registers." A developer could spin up a hundred thousand instances with just the click of a button.

Wix realized the financial risk that comes with liberal permissions to spin up infrastructure and hired Dvir in 2017. FinOps approaches infrastructure governance from a billing perspective and handles workloads already provisioned in the cloud. But at Wix's scale, where there are thousands of engineers, the FinOps approach stops working. "By the time you have a financial incident, it's too late and you didn't govern anything."

Dvir shifted the strategy to proactively preventing waste in the first place, by incorporating financial KPIs into engineering goals. In addition, Dvir built an internal platform called "InfraGod" which collects infrastructure data, integrates with Terraform, and enforces rules at the time of resource provisioning. Taking action at the time resources are provisioned rather than after the fact is "the difference between Finance and Financial Engineering."

Listen to this episode for a deep dive into the tactics that Dvir uses to run Financial Engineering at Wix, such as data collection, engineering post-mortems, monthly reports, and mandatory resource tagging.

Lars Kamp
Tobi Knaup

Tobi Knaup is the CEO and a co-founder at D2iQ, an enterprise Kubernetes platform. D2iQ combines the best open-source technology from the cloud-native technology landscape into a single Kubernetes solution. Customers can deploy this solution without worrying about the individual pieces they would otherwise need to assemble, maintain, and update.

In this episode, we discuss the shift to cloud-native infrastructure and how we are now seeing a new class of smart cloud-native applications emerge. Smart cloud-native applications include artificial intelligence (AI) components that leverage data from production applications. These new applications enable entirely new use cases in every industry. Examples are autonomous driving in automotive, medical imaging in healthcare, and fraud detection in banking or crypto trading.

To build smart cloud-native applications, companies need to build the infrastructure to train their AI models, put them into production, and build differentiated products. It is an entirely new type of workload, with very dynamic and elastic demand for compute and storage.

It turns out that Kubernetes, with its scheduling and orchestration capabilities, is a perfect fit to support workloads from smart cloud-native applications. Training models requires spinning up large amounts of compute to process data and then scaling back down. By putting model predictions into production, companies can lean on existing code pipeline workflows and monitoring.

This also means that instead of running two separate types of infrastructure, companies can consolidate and run their smart cloud-native applications on the same platforms as their production applications, which generate the data in the first place. The outcome for companies is highly differentiated digital products.

Listen to this episode to learn about Kubernetes, cloud-native architecture, and changes in organization and workflows that technology leaders need to adapt to deliver smart cloud-native applications.

Lars Kamp
Jon Edvald

Jon Edvald is the founder and CEO of Garden, an end-to-end cloud delivery platform that accelerates your development, testing, and CI/CD workflows.

In this conversation, Jon covers how the shift from monolithic applications to microservices has taken us from a single codebase to individual deliverables that are getting smaller and smaller. With the introduction of containers, an application now consists of many discrete components—which continue to get even smaller with the arrival of serverless. And where teams previously had to manage five to ten codebases, they are now dealing with hundreds or even thousands. Testing and deploying these different codebases has become a graph problem.

Beyond adopting containers and Kubernetes, the complexity of that graph of system components pushes the boundaries of existing DevOps tool chains. There is overhead for setup of each component in the graph, which becomes unmanageable with existing tools.

Garden solves this issue by factoring out things that are undifferentiated across different teams, allowing them to focus on their own business problems. Garden builds a directed graph of everything that needs to happen to transition from a bunch of Git repositories to a fully built, deployed, and tested system.

Listen to this episode to learn more about the industrialization of continuous integration (CI), infrastructure as code (with popular tools like Terraform and Pulumi), and how Garden helps developers ship more software faster.

Lars Kamp
Dieter Matzion

Companies build in the cloud for growth and speed. 📈

Engineering teams love building new things—so much so that cloud spend commonly becomes a major part of a company's profit and loss statement (P&L).

Cloud vendors have introduced pricing and discounting schemes to incentivize increased consumption and lock in long-term commitments from customers. Management gets involved at this point, but they often lack context and understanding of how cloud procurement works.

Forecasting cloud spend and aligning growth with infrastructure efficiency become important capabilities when you are about to sign a multi-million three-year contract with a cloud vendor. 💰

In this episode, Lars talks with Dieter Matzion, Senior Cloud Governance Engineer at Roku and long-time expert in cloud procurement and cloud financial operations. Before joining Roku, Dieter was an engineer at Google, Netflix, and Intuit, where he established infrastructure efficiency programs that combined cloud operations, analytics, and finance.