I'm facing a tough decision with one of our aging HPE Proliant servers that's frequently crashing with a Purple Screen of Death. Trying to avoid the costs of moving its workload to a different on-prem server, which would really strain our vCPU timesharing. I get that cloud solutions usually have a higher long-term cost, but I need solid numbers to make an argument for or against investing in new hardware. Given our small company's situation, I'm looking for ways to accurately measure our current CPU, RAM, and storage usage from VMware to help with cost modeling for cloud providers. I'm also considering using Grafana and Prometheus for metrics collection. Would cAdvisor be a good fit for getting VM and host metrics, or am I going off track? Any advice or suggestions would be greatly appreciated!
3 Answers
It sounds like you've got a solid plan. Just make sure you collect all the necessary metrics to compare against potential cloud providers. Keep in mind that often, engineering costs at data centers can be higher, which could impact your overall budget.
With almost 20 years of cloud experience under my belt, I can tell you that cloud billing often comes down to what you're actually using, especially with offerings like Functions as a Service. If you're mainly running your current workload 24/7, you'll want to find a cloud VM or instance that matches your requirements in terms of CPU and RAM. Once you select the instance type, just multiply the hourly rate by the number of hours per year. Starting small is a good strategy; you can always scale up later if needed. Just remember, cloud VMs tend to perform better running at 80-90% CPU utilization, especially during peak usage times.
Forgot to mention, I'm looking to offload around a hundred containers. They mostly sit idle until they get triggered by a process, with only a few services running around the clock. What do you think?
I'm open to any cloud provider as long as they have EU data centers. I've heard about the big players like AWS and GCP, but do you think there are any other good options for pricing or ease of use?
I've managed similar migrations before, and it really depends on your specific workload and how effectively you can leverage cloud resources. If you have questions, feel free to reach out, I'm here to help!
Could you clarify what you mean by 'engineering hours cost'? I'm new to this and trying to wrap my head around it.