Glad to hear that this project is still going! I thought for a bit that it was vaporware since I had not heard of any updates about it since your previous post. Do you have any sort of general timeline for the project so as to know when we can expect it to open to the public or have an open-beta, sort of?
The summer was a pretty hectic one. In addition to developing Cloudsmash, I also run a small datacenter called JeffColo.
In 2017, we decided to reorg and expand our datacenter into a larger space. We have had significant delays getting the new facility online. Before the move I suspected there would be problems, so in July I moved all the Oregon Cloudsmash gear down to Hurricane Electric in Fremont, CA.
The migration to Fremont was an excellent test for the resiliency of the mesh. Multiple peering points were dropped, vm's were migrated to alternate sites, hardware was packed up and driven to it's new home. Once back online, we established our new peering and began to migrate vm's back. It all worked perfectly, with zero customer downtime.
The move also allowed us to test using a virtualized equipment stack. This means that we have no physcial routers, firewalls, etc. Our servers connect to simple switches and rely on virtualized routers to establish upstream peering with HE.net. We received a 10 Gbps network drop, so we now get to validate networking performance beyond the gigabit threshold. Recently we have been able to get our virtual router instances to forward at the full 10 Gbps rate despite running on Westmere hardware from 2010.
Do you have any sort of general timeline for the project so as to know when we can expect it to open to the public or have an open-beta, sort of?
The previous beta round quota had been filled and I've been accepting feedback for the last several months. There have been small improvements here and there, mainly surrounding network performance and reliability.
Recently we began the process for hiring an architecture engineer. The selection process has been going well and it looks like we may have our first Cloudsmash developer coming onboard soon. I expect to have another consumer beta round in late Spring 2018 and hopefully will do a very limited alpha round for providers in mid-Summer 2018.
I have honestly never thought or seen anything like this before making decentralized VPSs and would love to see how such a technology could fare in a near-production environment. I do not have any specific host services in mind, but I do wish to see what can be accomplished with it.
I had not either and was frustrated that it didn't exist. It's certainly not vaporware, at this time it's a well tested stable platform that desperately needs a good web interface to interact with consumers and providers. Cloudsmash now has 14 pop's and several hypervisor locations, the network has experienced zero downtime since I brought it up in August 2016. It's kind of a living, breathing thing.