Hot on the heels of our transparency report update, we'd like to extend a HUGE thank you to the fine folks behind the Works on Arm initiative and Packet.com
We are happy to announce we have been awarded a slot in the Works on Arm initiative. We never, in our wildest dreams, thought we would be accepted.
We have worked very hard over the last 12 months to bring self-hosted clouds on arm mini-computers to the masses. Our work is focused on inexpensive, reliable boards with great documentation. We also are big on continuity of service if 'something happens.' All of our Docker images and docs are structured so that as many people as possible can deploy a self-hosted cloud on a Lollipop or remix it as they see fit. As part of the process we've been slowly adding more Docker images to our registry and recently we ended up at our capacity funding and hardware wise.
A fellow ActivityPub user pointed us in the direction of 'Works on Arm' when we started posting questions about alternatives to the Amazon AWS A1 (Arm) instance types (a very expensive service for us). We started our investigation shortly after finding out about the initiative, applied and were accepted. There were many e-mails, a phone call and a little bit of paperwork but we made it. 😉 (The process wasn't bad at all, honest!)
As part of our slot we received a 64 core, 128Gb RAM, 1Tb SSD server that is based around the HiSilicon Arm processors. To say we jumped about with joy would be understatement. 🤩
We spent the first 48 hours with the machine moving our Jenkins instance and all of our Arm build infrastructure over. We are happy, thrilled, and beyond thankful to the Works on Arm initiative for this opportunity. We can build all our Arm Docker images inside of 8 hours. Before this process took us at least 36 hours... That's a massive difference and gives us a lot of additional room to grow and add to our offerings. It also removes the need for external services like AWS to help with some of our 'big builds'.
If you're curious what our new hardware looks like we have the Munin data published here. Notable metrics are the RAM usage and CPU usage. We did a bit of work with the machine ahead of Munin coming online so the full CPU utilization we had (all 64 cores briefly) is missing from the graphs. That said, this is a very amazing piece of hardware and we look forward to adding more Docker images and integrations to ensure we max it out. 😀
We cannot thank Works on Arm, Packet, and especially Ed enough for this opportunity. Thank you!