Infosec.Exchange Update

Hello friends. just (quietly) celebrated its 5th birthday being part of the fediverse. For much of that time, the server had been running in a VPS. Over the years, I bought increasingly large VPSs from ssdnodes, and for the most part, the server would start out pretty quick, but then experience a significant performance decline.

A few weeks ago, Elon Musk in/famously made moves to purchase Twitter, which drove a bunch of people away from Twitter, looking for a new social media home. At the time, was already slow. Suddenly though, the server became unresponsive. The primary issue appeared to be I/O bandwidth to the disk, which is ironic given that disk I/O performance is the reason that one would use a VPS from a place called “SSD” nodes.

I was driving to my place at the beach this was unfolding. After I arrived and unpacked, I rebooted the server and freed up enough performance to ask the question: which provider should I move to. I ended up choosing the AX101 offering from Hetzner, which is a 16 core AMD Ryzen 5950x with 128GB of ECC DDR4 and 2x 3.9TB datacenter class NVME SSDs. Despite the server being in Germany and me being in the US, the performance is truly incredible for the price (~$100US/month).

It’s completely unfair for me to compare this Hetzner server to the one I bought from ssdnodes. The ssdnodes VPS had 48GB of memory, 12 vCPUs, and 720GB of NVME ssd space, but it cost me approximately $350 for 36 months (i.e., <$10/month). For the price, it’s not half bad, but in no way appropriate for the workload I am running.

Synching the data from the ssdnodes VPS to the new Hetzner server took about a day. I moved the database and key config files over to the new server and started it up, while the rest of the media continued to move over (about 250GB in total). I received a few complaints about missing media in that time, but after a day, all was good and the server went from about 10 second page loads to sub 1 second, even with many simultaneous users hitting it, all the while under the covers, the load on the system is negligible.

Oh, and an important note: upgrading to a new version of Mastodon would take about 30 minutes to complete on the ssdnodes server. Completing the same steps on the Hetnzer server takes about 2 minutes. The services restart so fast that it’s not noticeable to users (at least I don’t think it is).

Looking Forward

As you can see by reading this, I’ve added a WordPress installation to post announcements and the like in longer form than a Mastodon post permits. I’ll also have the maintenance maintenance/error page redirect to this WordPress instance to provide updates on what might be happening to cause an outage.

I have some ideas I’m working through on enhancing the security and monitoring of the server (as well as resiliency on the cheap) that I intend to post here for others to learn from/make fun of.

I’ve also set up a wiki ( that will contain details about the instance, rules, how-to’s, and possibly crowd-sourced info about security, if there’s interest.

Finally, I added a forum (https://infosec.exvhange/forum) for off-line discussions – I have no idea if that will be useful yet. Mostly did it because I had an itch to play with Simple Machines, which I haven’t done in about 10 years or more.

Happy tooting.






Leave a Reply

Your email address will not be published. Required fields are marked *