<p>I'm wanting to profile some stats on my live application to get an idea of peak memory usage for my various microservices. I'm looking through the runtime package to see what would be the easiest. Would I need to run func (*MemProfileRecord) InUseBytes in every part of the system, or can I just run ReadMemStats at the end of each request? The issue with that I'm seeing is that while it shows current memory usage and the total allocations, neither of those necessarily shows what I'm looking for. I'm trying to optimize the number of concurrent requests that I can process on a given instance size.</p>
<hr/>**评论：**<br/><br/>jerf: <pre><p>While those stats may not be precisely what you're looking for, it's likely that "precisely what you're looking for" would turn out to be very difficult to define if push came to shove. There's a similar question that gets asked by new Linux administrators, "How much memory is a given process using?", and it's really difficult to nail down what that question even means in the presence of shared memory. Those numbers are probably close enough that optimizing them will tend to do what you're looking to do.</p></pre>Ploobers: <pre><p>Thanks for the suggestion.</p></pre>
