4 ms·
PID is a damn sight more sophisticated than most datacenter dynamic capacity control algorithms - most autoscalers barely even qualify as ‘bang bang’ controller
by jameshart 5y ago
PID is a damn sight more sophisticated than most datacenter dynamic capacity control algorithms - most autoscalers barely even qualify as ‘bang bang’ controllers - they detect a need for more capacity, and add nodes at some artificially constrained rate until capacity is reached or they hit a max cluster size limit. Even rudimentary control theory is an improvement.
Of course the problem with applying PID to server capacity is that compute resources come in discrete chunks that are slow to bring online (‘computers’) rather than being a continually variable resource.
- lmilcin 5y agoI guess progress must be made in small steps... sigh...
- whatshisface 5y agoIf you have enough of anything it starts acting continuous.
- nine_k 5y agoI suspect that datacenters do not implement PID not because nobody there is aware of control theory. I suspect they tried and decided against it.
- oneplane 5y agoI think Hanlon's razor applies; most people in the datacenters have not heard of control theory whenever I interacted with them. The ones that did were usually from an academic background and might have had some side-track with robotics, industrial control software development or just plain theoretical studies. I suppose it's not as much that nobody wants it, or nobody wants to know about it, but there is so much else to be known that it might not 'fit' when assembling study materials or in-house learning systems. Maybe the best way to integrate control theory into the datacenter (or cloud) from an ops perspective would be starting out by getting some traction with control systems in general fist, just like a general software engineer might have had some software architecture and patterns for one or two semesters.
- j2kun 5y agoI work in datacenter optimization and we're aware of control theory.
- jcalvinowens 5y ago> Most autoscalers barely even qualify as ‘bang bang’ controllers - they detect a need for more capacity, and add nodes at some artificially constrained rate until capacity is reached or they hit a max cluster size limit. Even rudimentary control theory is an improvement. Why do you think PID is an improvement here? IMHO the bang bang approach is preferable because it behaves extremely predictably. Operationally, knowing how the system will react to extreme conditions might be more valuable than being a little more optimal.
- srean 5y agoParent's point is that they are worse than bang-bang controller. Its not very clear from your comment but are you claiming that PIDs are not predictable ?
- jcalvinowens 5y agoHuh? Parent was arguing in favor of it: "PID is a damn sight more sophisticated", "even rudimentary control theory is an improvement". Yes, PID upset is a real problem when things start going sideways, as they tend to occasionally in the real world.
- rualca 5y ago> Huh? Parent was arguing in favor of it: "PID is a damn sight more sophisticated", "even rudimentary control theory is an improvement". You clearly failed to read (and quote) what OP said, and therefore understand his point. Please pay attention to OP's point that PID is more sophisticated with the current state of affairs, which don't even qualify as bang bang controllers. The OP quite clearly and unambiguously stated that "Even rudimentary control theory is an improvement." Just because you feel the need to one-up other people's work that does not mean it is ok you distort and strawman the points they made.
- jameshart 5y agoMy point was more that the idea of applying control theory is an improvement over doing what amounts to bang-bang control without really thinking about it. Not that PID is better, not that bang-bang is better, just that.. learning from control theory is likely a fruitful direction to go in. Most autos along algorithms have heuristically hacked hysteresis built in to stop them oscillating between provisioning a node, dropping a node, provisioning, dropping. That kind of threshold tuning stuff is ripe for replacement with something more rigorous. PID control is an example of a rigorous way to solve that in continuous control. It likely doesn’t directly work for capacity, but control theory in general does have better answers for this kind of thing than just PID.