Oxide Computer Company’s cover photo
Oxide Computer Company

Oxide Computer Company

IT Services and IT Consulting

Emeryville, California 17,512 followers

Own Your Cloud

About us

Oxide is building a new kind of server. True rack-scale design, built with the innovations of cloud hyperscale technology, to make running on-premises compute infrastructure as easy as cloud.

Website
https://oxide.computer
Industry
IT Services and IT Consulting
Company size
51-200 employees
Headquarters
Emeryville, California
Type
Privately Held
Founded
2019

Employees at Oxide Computer Company

View 116 employees at Oxide Computer Company

or

By clicking Continue to join or sign in, you agree to LinkedIn’s User Agreement, Privacy Policy, and Cookie Policy.

See all employees

Locations

Updates

  • Oxide Computer Company reposted this

    I cannot tell you how stoked we at Oxide Computer Company are to be partnering with the Oakland Ballers! We are kindred spirits: East Bay-borne innovators always willing to experiment or otherwise challenge the status quo. And if you are our kind of nerd -- a baseball nerd, a computer nerd, or both -- sign up to join us on Saturday for our first Hack the Diamond hackathon! We will have Oxide folks on hand, including former professional athletes (and current computer nerds!) Zoltan Mesko and Justin Baum. Other Oxide nerds of note who may be sighted during the day include Adam Leventhal, David Pacheco, and Elliott Donlon. Bring your enthusiasm, your creative ideas, and your laptop -- and let's see what we can hack together!

    View organization page for Oakland Ballers

    3,187 followers

    Baseball has always rewarded feel, instinct, repetition and good coaching. But access to the best resources has never been evenly distributed. Too often, the technology that helps players improve, coaches teach and families stay connected is expensive, complicated or built for programs and organizations with bigger budgets. That’s what we want to help change. On June 13, we’re teaming up with Oxide Computer Company for Hack the Diamond, a one-day hardware and software hackathon. Goal: To bring builders together to create practical tools that make baseball more accessible, useful, affordable and fun at all levels. We’re looking for a diverse group to join us: designers, developers, coaches, baseball minds and curious builders from across the Bay to develop real solutions in a single day. If this sounds like you, we look forward to seeing you there: https://lnkd.in/egWZY6Tt

  • Oxide Computer Company reposted this

    Oxide Computer Company is actively updating the infrastructure running my website, my Bluesky PDS, and my Git server. All of those applications remain up through the update. Let's talk about why. First and foremost, the Oxide team has been hard at work delivering a self-service update experience in preparation for our live update experience. What's the difference? Self-service update: You click install and each compute sled stops its instances, updates, and starts its instances. Similar to a Kubernetes deployment rollout. Live update: You click install and each compute sled drains its workload, updates, and accepts workloads again. Similar to a Kubernetes node cordon, drain, and update. This isn't released yet but we're working on it! The update happening right now is self-service so every one of my instances running on Oxide will experience a reboot. How do we stay up in this case? Anti-affinity groups: I'm running a 6 node Kubernetes cluster, 3 control plane nodes and 3 worker nodes. I use anti-affinity groups to ensure the control plane nodes and the worker nodes are scheduled on different compute sleds. This guarantees that no two nodes will experience a reboot at the same time and lose quorum. Kubernetes guarantees that when nodes reboot the pods will be schedules on other healthy nodes. Floating IPs: The services running on Kubernetes are exposed externally via floating IPs attached to Kubernetes nodes. If the instance with the floating IP attached is rebooted, the Oxide cloud controller manager will reattach the floating IP to another healthy instance. Once we have an Oxide native LB this will be even more resilient! Distributed & Local Storage: Each control plane node boots from a distributed disk which means it can boot on any compute sled. This means I don't have to wait for a compute sled to finish updating to restart my instance. The worker nodes, however, are using local storage which is tied to a compute sled. Those will remain down while the compute sled updates. This is mitigated by running Longhorn on top of those worker nodes to provide distributed storage for the Kubernetes pods. Now my pods can schedule and maintain access to their data. Once we have our Oxide CSI plugin this will also be more resilient. So here I am, a former SRE, not breaking a sweat while my applications stay up while the underlying infrastructure is being completely updated. Not just OS updates either, but firmware, OS updates, and control plane software. The entire system is being updated while my applications continue to serve traffic. Oh, and I don't get charged a monthly bill for this. This is the future of on-premises computing.

    • The Oxide control plane web console showing an update to v20.0.0 is 15% complete.
  • Oxide Computer Company reposted this

    Keith's new fourth cloud model highlights one of Oxide's biggest strengths, you remain in control and the cloud boundary is clearly defined. No one can actually run VMware, Red Hat, or Nutanix alone, they don't include the underlying compute, storage, and network infrastructure. Similarly, many customers don't run only Oxide, because we've drawn a clear boundary at the VM layer, containers and additional services are added on top. Our case study with SUSE and Switch is a great example- Oxide provides best in class infrastructure with co-designed hardware and software that when combined with Rancher creates a fourth cloud that keeps you in the driver's seat and smooths out the typical pain points of legacy commodity servers, storage, and networking gear that were never designed to run a cloud. Similarly, OpenShift or custom cloud control planes are excellent choices to run on Oxide thanks to our open source, secure, and well-documented platform. Think of Oxide as shaped like EC2, EBS, and VPC but open-source, on-prem, with the best hardware you've ever had access to, and fully under your control. Your platform team starts day one on a substrate that's already API-driven end to end, instead of assembling and supporting your own, it's your team's massive head start and advantage over the field.

    The most complete on-prem AI platform in my new assessment is also the most captive. Completeness and control turn out to be different axes — and almost every infrastructure evaluation measures only the first one. I scored VMware Cloud Foundation, Red Hat OpenShift, Nutanix, and Oxide Computer Company against the Fourth Cloud model on both. The authority numbers are where the surprises live: two of the four are ~65% locked to the vendor, one is 85% portable, and the one everybody underestimates on capability is the most authority-preserving of the group. Full breakdown — including the corrections all four vendors pushed back on — in the article. 👇

  • Oxide Computer Company reposted this

    External subnets are here! Bind directly to external IPs from inside your Oxide Computer Company instances. This unlocks new networking patterns — including one Kubernetes users will care about. Previously, external IPs allowed instances to be reachable from outside their VPC, but those IPs were transparent to the guest OS. That meant applications inside the instance couldn’t bind directly to them. With external subnets, an instance can see and bind to addresses from an attached external subnet just like any other interface address. Use cases this enables: - Multiple external load balancer endpoints. - Skipping the Kubernetes overlay network. - Binding the same port on different external IP addresses. See it in action in the video below.

  • Oxide Computer Company reposted this

    Two of my biggest passions, sports and technology, are coming together in the best possible way. On June 13, the Oakland Ballers and Oxide Computer Company are coming together to host Hack the Diamond, a hackathon focused on building tools that expand access to baseball technology for everyone. Technology shouldn't just be available at the highest levels of the game or for those with the right resources. This hackathon's goal is to bring builders together to explore ideas that can make baseball technology more accessible and useful for players, families, teams, coaches and communities — everyone. If you live in the Bay and this is your lane, we’d love to have you join us.

    View organization page for Oakland Ballers

    3,187 followers

    Baseball has always rewarded feel, instinct, repetition and good coaching. But access to the best resources has never been evenly distributed. Too often, the technology that helps players improve, coaches teach and families stay connected is expensive, complicated or built for programs and organizations with bigger budgets. That’s what we want to help change. On June 13, we’re teaming up with Oxide Computer Company for Hack the Diamond, a one-day hardware and software hackathon. Goal: To bring builders together to create practical tools that make baseball more accessible, useful, affordable and fun at all levels. We’re looking for a diverse group to join us: designers, developers, coaches, baseball minds and curious builders from across the Bay to develop real solutions in a single day. If this sounds like you, we look forward to seeing you there: https://lnkd.in/egWZY6Tt

Similar pages

Browse jobs

Funding