NixOS and Metal, Back in Action
If you dig NixOS and you love Metal, then today is a happy day!
Thanks to a collaboration with Determinate Systems, NixOS 21.05 "Okapi" is now available as an official OS for Equinix Metal. That means you're just one click (or API call) away from enjoying your favorite declarative operating system on some sweet bare metal.
  • NixOS 21.05 is available on just about all configs.
  • Notable exception: c3.small.x86 is not yet ready, but coming very soon.
  • Yes, this includes x86 and Arm configs (example, c3.large.arm). Woot!
While adding a new OS to our list sounds simple, a
of great "under the hood" work has taken place to make it easier and more reliable for us to validate and then deliver images across our configs.
Huge props to our Delivery Engineering team to Graham Christensen and Determinate Systems for working with us and being the first one through the gate.
Time for some juicy Arm goodness - thanks Ampere!
Today we're happy to announce our latest Arm-based server (the c3.large.arm) which is based on Ampere's 80-core Altra processor. Want a closeup? Here's a picture!
We've worked with a long line of Arm machines over the years...from the original 96-core Cavium ThunderX and the follow-up ThunderX2, to the Qualcomm Centriq, various boxes from Huawei, and more recently the Ampere eMag.
Suffice it to say that we've come a long way and the c3.large.arm stands heads and shoulders above previous generations. Altra was custom designed by Ampere for cloud-native workloads, which translates into lots of scale and consistency. Here are two stats to help you see what we mean:
  • 80 physical cores in a single socket design.
  • Consistent scaling at 2.8 GHz per core. Run 'em all full out!
Our c3.large.arm is available today in Dallas and Washington DC, with more sites and inventory coming soon. Enjoy!
OSS Alert! Declarative Kubernetes cluster management w/ Tinkerbell
Today we’re excited to announce the Technical preview of Cluster API Provider Tinkerbell.
With this release, it is now possible to provision, scale, and upgrade Kubernetes clusters on hardware managed by Tinkerbell in a declarative fashion using Cluster API.
Check out the project documentation to learn more.
Equinix Labs - our new home for sharing code
We're excited to share the creation of Equinix Labs, a place for Equinix staff to collaborate on code with each other and with customers.
It's a GitHub Org that we expect to use as a home for projects that need an official Equinix GitHub URL but aren’t part of any formal product codebase. We are already seeing it used for Terraform modules, small one-off tools and scripts, and code samples that accompany technical guides.
Equinix Labs is in its early stages and we are actively welcoming Equinix staff and customers to use it as a collaborative space - its future will be shaped by those who engage with it!
To find out more, head over to: or our Slack channel.
Dr. Heinz Doofenshmirtz, with Candace, in the series finale of Phineas and Ferb. Image courtesy of Disney XD
New Kill Bill Billing Engine Live
As you may have read in this recent blog post by Pierre-Alexandre Meyer, we're hard at work building a next-generation billing platform to power Equinix Metal.
Billing systems aren't the reason you choose an infrastructure provider, but they form the basis of the entire commercial experience, especially in usage-based models. With full credit to Twilio's CEO Jeff Lawson, that's why you "build" instead of "buy" and why we've invested in Kill Bill (an open source project whose maintainers work at Equinix) to help us get there.
What's it mean to you?
Starting with the October billing run we've moved a portion of our customers to this new system, with the final transition in early 2022. If you're on the new system here's what you can expect;
  • Instead of receiving multiple invoices (one for each of your projects) listing all individual line items, you will receive one simplified, consolidated invoice.
  • This new invoice will present your services based on their subscription type (Usage, Committed, and One-Time fees).
To support the next stage of our growth, we've also introduced a more powerful "Billing Account" structure under the hood (more on that in the near future). We'll expose this once our transition is complete to help make managing your subscriptions and analyzing monthly billing easier and more powerful.
Any questions? Please reach out and we'll be happy to assist!
São Paulo Metro Live!
It is with great pleasure that we can (finally!) pull the wraps off our São Paulo metro, our 18th since last summer, and our 13th new metro this calendar year.
Our operations team deserves
so much
credit for conquering a variety of challenges, not the least of which was the ongoing Covid-19 pandemic, which has hit our friends and family in Brazil quite hard. We even wrote down some lessons learned along the way.
With the SP Metro live, you can now deploy both m3.large and c3.medium on-demand (with s3.xlarge coming in soon). Everything, of course, is also available via reservation.
One thing to note is that, like other clouds, we have higher costs in Brazil than in places like the US or Europe. You'll notice that on-demand pricing is 50% higher than our baseline.
Any questions? Want to go with us to Brazil? Reach out and let's have a chat!
Metal Gateway
Want to simplify network connectivity for subnet traffic on Metal? Our new “Metal Gateway” feature automatically provisions a Metro-wide gateway for servers deployed from a new or unused Public or Private IPv4 subnet (/25 to /29) block. Magic!
Take a look at the docs to get up to speed.
Also, this feels like the right picture, doesn't it?
New "Dedicated Tunnel" Networking Mode
This feature is targeted at those of you with large non-Metal environments.
Let’s say you have a few hundred VLANs on the other side. With Dedicated Tunnel Mode, instead of configuring each of those in our portal, you can set up a single “dedicated tunnel”, tag the VLANs, and et voila - you’re in business. In other words, you can leverage a dedicated port as an 802.1q tunnel. Also known as QinQ!
Interested? Check out the docs for all the details.
Also, there are so many cool tunnel pictures on Unsplash.
halil-ibrahim-cetinkaya-SL7s6KYdGuQ-unsplash (1)
Update Alert: Cloud Provider for Equinix Metal
Howdy folks —
It's been a few months since we cut a release for one of our favorite Kubernetes tools: the one and only Cloud Provider for Equinix Metal (or the artist formerly known as Cloud Controller Manager)! As you'll see from the list below, we've crammed some good stuff into the latest version.
  • Bugfixes, because, ya know.
  • Permission fixes
  • More annotations on kubernetes nodes for private network ranges
  • Load balancer selector configuration renamed
  • Enhanced BGP peer documentation
  • Helm chart enhancements and fixes
  • Ability to use existing services for control plane Elastic IP port and healthcheck port
A tasty bunch of integration updates
We've been cooking up some tasty improvements to our CLI, SDKs, and integrations. Grab a plate and load up at the buffet! :-)
This version adds a new
packet env
command for copying your configured API token into your shell environment. Release notes.
Introduces Metros to the client. Metros can be used instead of facilities when creating devices, IP addresses, and VLANs. This release also fixes a bug in handling new resources that are assigned to metros but do not include facility details. Release notes.
We've added features for getting the bandwidth from your instances, determining spot market prices, and understanding ports from virtual circuits. A noteworthy bug fix corrects the pagination of hardware reservations. Release notes.
This release adds support for metros and updates the default location, OS, and plan so that all Equinix Metal users can take advantage of the defaults. The termination time for instances was not being honored previously, so we've fixed it. You can specify it in relative terms like
--metal-termination-time="60 minutes"
. This server will self-destruct (but not really). Release notes.
Our latest Terraform release deprecates
resources alongside the end of life of our Block Storage service on June 1, 2021. Users have been notified directly about this (more details).
  • On the topic of deprecations, if you received a warning about facilities being deprecated, fear not. This was a bug that we've corrected!
  • v2.1.0 also brings a number of new features including data sources with support for VLANs, IP reservations, ports, and hardware reservations.
  • Finally, resources have been added to bring support for Equinix Metal-Fabric Connections and Virtual Circuits. Release notes.
Load More