Today, it’s all about apps and digital services – not just to keep the business running – but also to grow and compete even more aggressively. But modern cloud native development can be complicated with the lack of true pipeline integration across tools for application platforms, security, and networking for an app-first world. And this is exactly what inspired us to create ScaleX.
Watch our prestigious lineup of presenters who have shared their experiences and knowledge around building for scalability and reliability and what that means for the modern cloud native developer. This second edition of ScaleX, hosted by Cisco, featured a full day of sessions, deep dives, and use cases from people just like you…the people building, operating, and maintaining secure, reliable cloud native systems at scale.
Vijoy Pandey, Vice President of Engineering, Emerging Technologies, and Incubations at Cisco welcomed both attendees and presenters, while providing an overview of the day’s activities, and discussed the impetus behind the creation of the ScaleX event as well as the wealth of expertise our presenters bring to the community by sharing their real-world experiences of building and operating cloud native systems at scale.
Vice President, Engineering for Emerging Technologies and Incubations, Cisco
Since the launch of Bitcoin in 2009, the world has been captivated by the vision of an Internet-native currency that’s flexible, borderless, and evolves with the times. However, both Bitcoin and its most popular successor, Ethereum, have fallen short of that grand vision. The reason? Inability to scale.
Enter Solana, the world’s fastest (400ms block times) and cheapest ($0.01 USD per transaction) decentralized blockchain. These factors enable crypto apps to achieve performance on par with that of traditional financial applications. Learn how Solana’s key innovation, Proof of History, enables unprecedented scale for decentralized finance.
Watch an example of how the concepts of SD-WAN, SASE, and Cloud Native WAN are implemented on ngena’s global connectivity platform. A scalable SD-WAN as a Service together with comprehensive Network as a Service capabilities will provide the basis to further improve application performance at scale while providing a clear interface between Kubernetes and SD-WAN as well as an end-to-end orchestration approach.
Founder and Managing Director, ngena
In this follow-up to our presentation at the first ScaleX event this past November, we provided an update on the evolution taking place right now at BBVA to go from a monolithic architecture to one based on microservices. We emphasized the differential aspects of our architecture and how it has been implemented using Kubernetes and other CNCF projects.
Head of PaaS, BBVA
Senior Expert Architect, BBVA
Liqo.io is a novel open-source project that can dynamically create a “big cluster” by transparently aggregating multiple existing Kubernetes clusters. A point-to-point protocol dynamically negotiates all the details of the interconnection (e.g. network parameters, resource quota), while seamless scheduling is obtained via a pod offloading mechanism, using a virtual kubelet to masquerade an entire cluster as a single node (called “big node”). Finally, scalability is achieved by a two-level scheduling and selective event forwarding, all running on vanilla K8s.
While developing this project, we quickly realized how this paradigm could be used to create a cluster of clusters, a two-level hierarchical architecture that can potentially guarantee an unprecedented degree of scalability in Kubernetes. However, this would introduce new challenges about how the network is extended across all clusters, how a scheduler can decide the optimal location for the running (micro)services, how information is stored (and cleverly duplicated upon necessity) in the etcd distributed database for state synchronization across clusters.
Watch how this talk briefly introduces the Liqo project, its “cluster of clusters” model, and it highlights the main challenges (and possible solutions) that have to be addressed to drastically scale the current Kubernetes technology to potentially millions of connected nodes, finally realizing the ultimate Liqo vision, “Building your endless Kubernetes ocean.”
Associate Professor, Department of Control and Computer Engineering, Technical University of Turin
Fundamental Architectural choices made early in an endeavor end up defining where the limits of scale of a system are. Watch this session as we look at some of those decisions through the lens of systems that are small scale (network controllers), medium-scale (K8s), and large scale (global DNS) and highlight lessons that can be learned from each for designing for scale.
Distinguished Engineer, Cisco
Large-scale migrations are difficult because they are often slow, complicated, and a draw on resources. While the infrastructure team sees the value in migrating to the latest Kubernetes version or ingress-controller to ensure the reliability of the system, a developer team might not be as enthusiastic. They have different demands on their time like delivering features or fixing bugs. The infrastructure team at Under Armour realized that if we wanted to scale and move quickly, we had to make our migrations as transparent to development teams as possible. Watch and learn how Under Armour uses Kubernetes internals to manage large-scale migrations without developer involvement.
Manager, Infrastructure Engineering,
The industry is shifting! There is a gap between our current threat models and the new heterogeneous and dynamic networking and application stack. The gap between our assumptions and reality will open up new attack vectors which will require a new security paradigm to combat.
Watch as we discuss the current assumptions, what needs to change, and promising architectural patterns which may help us defend our new massively distributed infrastructures!
Director of Software Engineering, doc.ai
In the cloud world, it’s incredibly easy to set “autoscale” for compute, memory and storage. You drop an application across a few clusters, set up autoscaling, and let the app efficiently use as much CPU and memory as it needs. Until now, you weren’t able to scale bandwidth accordingly across non-co-located clusters. Watch PacketFabric show how in a joint work with Cisco where they introduced the concept of Wide Area Network autoscaling and how application deployment can now directly autoscale the inter-cluster network dynamically. With Wide Area Network autoscaling the WAN is delivered just like any other cloud resource.
Co-Founder and SVP Product and Engineering, PacketFabric