Microsoft's quick-start virtual machines; Nutanix eyes VMware

Welcome to Runtime! Today: a new open-source project from Microsoft aims to marry serverless computing speed with virtual machine security; Nutanix tightens its links to AWS, and the quote of the week.

(Was this email forwarded to you? Sign up here to get Runtime each week.)

Ship it

Shine a light: Virtual machines have been the atomic unit of enterprise computing for decades, but they're a lot slower than developers working on real-time applications would prefer. Microsoft released an interesting new open-source project this week called Hyperlight that promises to improve the launch time of applications running on virtual machines without having to give up the security benefits of using a hypervisor.

Hyperlight is a Rust library that "you can use to execute small, embedded functions using hypervisor-based protection for each function call at scale," Microsoft said in a blog post. It's not as fast as WebAssembly, but it can spin up VMs in "one to two milliseconds" according to Microsoft and "with Hyperlight we can take those same runtimes and place them inside of a VM that provides additional protection in the event of a sandbox escape."

Sort it out: Security professionals have a lot of tools at their disposal to track threats and assess vulnerabilities, which sounds great until you understand how many alerts those tools send out on a given day. Large-language models could help security teams triage those alerts, and Fortinet rolled out two services this week in pursuit of that goal.

Fortinet's FortAI service now works with its FortiNDR Cloud service, which analyzes network activity, and Lacework FortiCNAPP, which protects cloud applications and was acquired by Fortinet earlier this year. Like every sector of enterprise tech, security vendors are throwing generative AI technology into basically everything they ship in hopes of sparking new interest in their services, but defenders are excited about its potential.

Yeah we do cloud: While VMware seems much more focused on its core private cloud data center customers these days under Broadcom, rival Nutanix is cozying up to AWS. The two companies announced an expanded partnership this week that includes shipping a new version of Nutanix Cloud Clusters that will help Nutanix customers manage applications and data across their Nutanix data centers and AWS.

Nutanix will also offer discounts for customers migrating away from VMware on AWS, which was a key part of VMware's product strategy during the growth of cloud computing until priorities changed after the Broadcom acquisition. While there's been a lot of talk about cloud repatriation over the past few years, companies are clearly launching generative AI projects on the cloud providers and the ones who still have substantial on-premises infrastructure will need something to help manage workloads across both environments.

Platform diving: Upbound got ahead of what appears (from my inbox at least) to be dozens of launches scheduled for next week at KubeCon with the release of two new features for its developer platform. Based around Kubernetes and Crossplane, an internally developed open-source project, Upbound helps users deploy and manage cloud resources when building their own applications.

The platform now supports VS Code, the widely used software development editor, and allows developers to create "lightweight, testing-focused control planes" for "real-time debugging and service validation," according to a press release. As platform engineering continues to emerge as the next evolution of the DevOps philosophy, companies are looking for developer-friendly ways to manage complex cloud operations.

Ditch the sidecar: When the open-source Istio service mesh was first developed it relied on a blueprint that contained a "sidecar" data plane, which delivered several powerful security benefits but was costly and cumbersome to operate. This week the Istio project announced that an alternative to the sidecar, known as "ambient mode," is now generally available.

"As a community, we designed ambient mesh from the ground up to tackle these problems," wrote Solo.io's Lin Sun, a member of the project's technical committee, in a blog post. The ambient concept was first developed in 2022, but with this week's announcement the project feels comfortable that it is ready for production use.

Stat of the week

Like many companies over the last decade, Amazon built a lot of its internal data stack around Apache Spark. But when it switched data-compaction tasks to Ray — the open-source data project developed by the founders of Anyscale — Amazon saw an 82% improvement in performance, which saved a ton of money, according to a talk presented last week at All Things Open by Amazon's Patrick Ames.

Quote of the week

“Trump will bulldoze you if you are too sycophantic. Collective action is needed to avoid being victimized.” — Jeffrey Sonnenfeld, a professor at Yale School of Management, explaining to the New York Times why nearly every major tech executive delivered essentially the same message of congratulations this week to the president-elect after Tuesday's election.

The Runtime roundup

Amazon is in talks to put more money into Anthropic, but only if the model startup uses "a large number" of AWS's custom AI chips, according to The Information.

Venture capitalists are trying really hard to throw money at Anysphere, the company behind the Cursor coding assistant, according to TechCrunch.

Thanks for reading — see you Tuesday!