tamuma
Member OP
AWS spot hit $0.0032/hr for t4g.nano and I felt my soul leave my body because my whole business model is charging $3/mo for smaller.
AWS dropped spot prices again, feeling conflicted
Poggers for ur wallet, vibe check: r u the product now or the customr lol. Literally hyperscalrs gonna hyperscalr, bettr off pivoting to support as a s3rvice. Based if u do, cringe if u dont. I ran my minecraft srvr on spot for 8 cents last month, electricity costs mor than that.
IPv4, IRC, and irssi — fight me
Brooo thats wild. I got a sevrer form Contabo for liek 2 buck and its fire but aws spot is cheat codes. My hosintg budget is literally zero this month. Cant complian tho!
Vive la résistance... électrique
Pro tip: if you are worried about margin compression, diversify into workloads that hyperscalers do not optimize for.
- Set up Always Free tier ARM instances on Oracle Cloud as your control plane.
- Use Contabo or OVHcloud for the actual customer payloads.
- Keep a spot fleet for bursty batch jobs only.
Heads up: Oracle free tier requires patience with capacity availability, but the price is unbeatable for permanent infrastructure.
licensing is a suggestion
giorgiauc
Member
Electricity costs in my region went up 40% year over year, so any pricing discussion needs that context. Optimization is not about the headline rate, it is about what you can run per watt. Some of us have workloads where the compute is actually the cheap part.
Hyperscalers are not your competition, they are the reason IPv4 lease rates hit $60 per IP last week. The low end is being squeezed from both sides.
LichunLars
Member
Route via Vultr adds 18ms from my test box. Route via Contabo adds 9ms. Hyperscaler backbone adds 3ms. Numbers do not care about your business model.
vultr.com: 18.2ms avg
contabo.com: 9.1ms avg
aws-ec2.example.com: 3.4ms avgCustomers who need low latency will pay for it. The rest will chase spot prices forever.
1ms or I don't want it