I'm a recent college grad working part-time at a small startup that has been mainly focused on marketing, but we're now branching into SaaS. Currently, I'm managing infrastructure for a team of two developers who are working on our HR management system, which so far has only been used internally. We've just received news that we might have an enterprise client with about 1000 employees interested in using our system, and my manager is asking for a monthly cost estimate for hosting this app for them.
Previously, I gave an estimate for about 30 users, but the company is now looking for a specific number that can support a much larger user base. I've been trying to figure out the right capacity we will need, especially since our app shouldn't scale down to zero. I'm aware that determining the utilization is tricky without solid usage data, and my only reference is one month of 30 users. Currently, we run a serverless setup that has been cost-effective, but my manager isn't keen on that approach for the enterprise deployment.
How should I approach estimating the required infrastructure and the associated costs? I'm especially looking for a methodical and data-driven approach that would strengthen my estimates, especially since I've encountered some pushback from management and there's interest in AI recommendations which I feel may not apply accurately. Any advice will be much appreciated!
4 Answers
You should definitely start with the Azure Pricing Calculator! It’s a great tool that helps you gather insights into what your specific configuration will cost based on usage patterns and services you choose. Make sure to outline all the Azure services you're using — that info will help you get a clearer picture of your costs.
It sounds like you're dealing with quite the challenge. Over-provisioning for your first enterprise client can be a safer bet if you’re uncertain about traffic and usage. Consider using slightly more powerful VMs than you initially think you'll need. This will help avoid any crashes or downtime that could occur if the system can’t handle the load. It's better to be cautious early on. And don't forget to set up monitoring with Application Insights—it'll help you identify issues before they escalate into bigger problems!
That’s a smart approach! Having diagnostics in place will absolutely help in figuring out how the app performs as user numbers grow.
Take a close look at what services you’re using in Azure and what specifications you anticipate needing for those 1000 users. You should assess if you can run multiple users on a single unit of compute to optimize costs. Also, running a load test could help you understand better how the system might behave under higher user loads, giving you a more accurate idea of what to expect in terms of performance and cost.
Honestly, it really depends on the expected load and usage patterns. If your app can handle more users with existing resources, there might be some savings. But I'd advise you to map out your maximum allowable downtime as well, since HR data security is critical. Management should understand the need for robust infrastructure, so if they’re relying on AI too much, push back on that and focus on solid data instead. Do your homework on usage patterns before finalizing those estimations!
Thanks for the pointers! I get that scaling accurately means I need to dive deeper into usage, and I'll definitely emphasize the importance of situational analysis with my managers.
Also, remember to include metrics when justifying your budget! They'll appreciate the groundwork you've laid out.
Great idea! I’ll incorporate Application Insights to track everything effectively. That should provide the data I need for my capacity planning.