Member OP
Honest comparison of budget dedi vs high-end vps
Need real data. Running i/o bound workload, lots of small random reads, not much cpu. Considering fiberdole bare metal vs GreenCloudVPS 8gb plan. Anyone got actual fio numbers? Not synthetic, real world. BGP and v4 only, v6 broken at my upstream. Peering
Now it's all "nvme" this and "someone else's computer" that. Ran a VAX once with 64MB RAM and it served 500 users. Your "high-end vps" is a fraction of a machine you don't control. systemd probably thrashing your I/o scheduler anyway.
Back in 2009 I had a dedicated atom at OVHcloud, €15/month, single sata drive. They don't make them like that anymore. Thing is, for my sqlite-heavy app, that old spinner beat a modern vps on GreenCloudVPS for random reads. Something about not sharing the controller. I still have the benchmarks somewhere, let me dig.
Edge. Pop everything to edge, origin stays calm. I ran same workload on bytearray bare metal vs HostHatch vps, cache hit ratio told the whole story. It's 12 now. Pop your static, let the origin breathe. Miss on first request, hit forever after. Latency without pop is just origin sadness.
Member
Enterprise-grade deployments require redundant feeds and N+1 at minimum. Our facility offers both bare metal and virtualized solutions with redundant power and cooling. For I/o bound applications we recommend speaking with sales about our enterprise-grade storage tier. Location is undisclosed for security.
Member
Hey folks! Ran this exact test last month, gonna share my screwup. Got a cheap dedi at Leaseweb and a "high-end" vps at CloudCone. fio random read: dedi was 2x faster. BUT I had the vps on wrong plan, their "storage optimized" tier crushed the dedi. I was super stoked to be wrong. Check the fine print, folks.
