LinkedIn·Wednesday, 12 August 2026·15d ago

2.5 hours of job backlog. Gone in 8 minutes. Total cost: $0.002. Two tenths of a cent. I checked it three times because I did not believe…

Ronald Dehuysser

Founder and Developer at JobRunr

2.5 hours of job backlog. Gone in 8 minutes. Total cost: $0.002. Two tenths of a cent. I checked it three times because I did not believe the number. What happened: we dumped 150 jobs of one minute each onto a cluster with a single worker. A MacBook Pro M4 was technically the production cluster that day 😅 Then JobRunr looked at the queue, decided it needed help, and rented a spot instance in our own AWS account. 6 seconds to decide. 75 seconds until the first spot node was processing jobs. Under 8 minutes for the whole backlog. 14 minutes attached, then it gave the machine back. $0.0019 in total. Let me be precise about what this is, because it is easy to oversell. Your existing servers keep running exactly like today. Spot adds burst workers when the queue backs up, and retires them when it drains. And let me be equally precise about the saving. The discount is AWS spot pricing, not our magic: 62 to 69% below on-demand for the identical instance type. What we built is the part that captures it automatically, inside your own account, so nobody has to watch a queue at midnight. The whole test day, 270 jobs and 4.6 hours of compute, cost $0.012. Do you want your manager to give you a promotion because you saved your company 60% on cloud costs? DM me, and I'll help you out.