Altivox's Datacenter Migration: How it went

Hi everyone, Cam here! We've recently faced a lot of issues with our old datacenter provider in Springfield, MA to the point that we have moved out of the facility entirely. For those of you unaware, a migration at this scale is extremely difficult. Here's how we did it:

Why we migrated

Before Altivox was founded, it was originally known as StaticNode, where the CTO of the company had decided to purchase a rack through this datacenter provider to run StaticNode out of. Before it merged with SaharaCompute to form Altivox, I hadn't really been interested in the location, but eventually came to terms with it and formed Altivox with David.

I personally wasn't a huge fan of the location, but it was a paid-for rack with servers already in it and there hadn't been any issues yet, so I rolled with it and worked from there with David.

Then the issues started...

Roughly around July 19th, 2025 we had gotten an email stating there would be power upgrades that would not affect customers, so we did not notify any of our clients (which was a huge mistake on our end). Come the maintenance day at 9:02 AM EST, we were notified of a complete outage in our entire rack, and were also unable to enter the facility.

We were in shock. We called their support who were seemingly unaware of this maintenance, and after 8 hours of downtime, at 4:00 PM, we found out that our services had returned. We made announcements in both our billing panel, as well as our community Discord where we explained the situation to all of our clients:

As you can see, we had no more plans of staying in this datacenter and had started planning a move, but this is not where the problems end.

Cogent: Part 2

Not even 10 days later, we had yet another issue with Cogent Communications severing all of our services and requiring us to pay a $4600 bill for "IP Abuse." They gave us no notice and forwarded no reports to us. We were locked out the DC with our machines, as well as our clients' machines unable to be accessed.

Ultimately we had to pay the bill and there was nothing we could do about it. It had seriously affected our morale and our ability to grow our business and, as of writing this, we are still extremely bootstrapped and low on company funding. (But hey! Everything's been moved and things are sailing smoothly now!)

How we moved

Well, now here comes the hard logistical part. See, here at Altivox we have a couple of different services: Virtual Machines, Game Servers, Bare Metal, Colocation, etc, so it wasn't a simple move.

• Entirely new network stack (Juniper -> Arista core)
• Migrate Bare Metal and Colocation clients first
• Migrate Virtual Machines once network is up

Once we had this plan in place, we had our new rack deployed at 3500 NW Boca Raton Blvd through 365 Datacenters (who managed to deploy the rack + power and internet within 2 business days) where a few buddies and I installed all the initial gear.

Now we had a few issues getting things operational. First we dealt with a weird problem, which a lot of people tried helping us with to no avail:

Now, what was the solution you may ask? Auto-negotiation of course. For those of you who do networking, who've worked for a NOC, or do anything of the sorts, I'm sure you can relate. The line from the DC is a redundant 1GE backup line that we're using for fallback transit, whilst we use Cogent + FL-IX and (hopefully soon) Breezeline for main transit.

Our router was trying to negotiate to 10g, whilst the DC's core switch was trying to negotiate at 1g, causing an issue for getting the link state up. This is why fiber tests worked between each point, but ultimately failed in the end.

After solving this, it was all about our internal configuration. Our NOC did an excellent job with getting our network running up to speed, we now feature up to 320G internal traffic and downstream capacity, as well as a future upstream blend of upwards of 30-40G.

We do still have a few issues to iron out, such as our core router (an Ubiquiti EdgeRouter Infinity) not supporting Source IP modification in its kernel making it extremely hard to do GRE Tunnels as well as join IXPs. Our solution: Juniper MX250 shenanigans.

Conclusion

It has been a rough ride recently but we've made it. Altivox is now running higher end hardware, a better network and a much better location.

We could not have done it without the support of the community, as well as our amazing staff team helping us along the way. We look forward to the future and hope it brings nothing but greatness for us, and whatever you may be hosting on our platform.