An independent regional operator running its own EPYC hardware in Warsaw, Vilnius and Singapore, with local peering rather than backhauling traffic to the big Western European exchanges. Self-managed KVM instances with NVMe storage, hourly billing and a modest but functional API. Strong price-per-core in its home regions, thin coverage everywhere else.
Share your experience to help others choose the right provider.
12ms from Vilnius, and they peer locally instead of hauling everything to Frankfurt and back. Hardware is EPYC and the NVMe is quick - I get about 90k random write IOPS on the mid plan. If your users are in CEE this is a genuinely good option that almost nobody talks about.
Short blips, 20 to 90 seconds, frequent enough that my uptime monitor is permanently noisy. Never long enough to really hurt but it chips away at confidence. They acknowledged an upstream transit problem in Warsaw and said it was fixed; it is better than it was, but it isn't gone.
We run three build agents here and the cost is roughly 40% of what the bigger names quoted for comparable dedicated cores. Provisioning is a couple of minutes, hourly billing works properly, and the API is small but does everything I need. Routing to Australia is mediocre, worth checking if that matters to you.
Median first reply is somewhere around five hours, which is not good. But when a human does answer it's an actual engineer who reads your logs and doesn't send a canned article. I'd take slow-and-correct over fast-and-useless, though ideally I wouldn't have to choose.
Advertised dual stack, delivered a /64 that didn't route to a large chunk of the internet. It took repeated tickets and roughly three weeks to sort out, with no acknowledgement on the status page. Everything else has been solid and the pricing is good, but test what you actually need before you prepay for a year.
12ms from Vilnius, and they peer locally instead of hauling everything to Frankfurt and back. Hardware is EPYC and the NVMe is quick - I get about 90k random write IOPS on the mid plan. If your users are in CEE this is a genuinely good option that almost nobody talks about.
Short blips, 20 to 90 seconds, frequent enough that my uptime monitor is permanently noisy. Never long enough to really hurt but it chips away at confidence. They acknowledged an upstream transit problem in Warsaw and said it was fixed; it is better than it was, but it isn't gone.
We run three build agents here and the cost is roughly 40% of what the bigger names quoted for comparable dedicated cores. Provisioning is a couple of minutes, hourly billing works properly, and the API is small but does everything I need. Routing to Australia is mediocre, worth checking if that matters to you.
Median first reply is somewhere around five hours, which is not good. But when a human does answer it's an actual engineer who reads your logs and doesn't send a canned article. I'd take slow-and-correct over fast-and-useless, though ideally I wouldn't have to choose.
Advertised dual stack, delivered a /64 that didn't route to a large chunk of the internet. It took repeated tickets and roughly three weeks to sort out, with no acknowledgement on the status page. Everything else has been solid and the pricing is good, but test what you actually need before you prepay for a year.