Feedback Control for Computer Systems: Introducing Control Theory to Enterprise Programmers

I forgot almost everything I studied when I was in college. This book reminds me what I have learned. Also since it passed almost 25 years, I learned quite some new things. I am a mechanical not a computer engineer but this is really good start. Very good choice!

This should be required reading for any software architect or developer that's working on large scale distributed systems, microservices, cloud computing... anything where correct behavior depends on corrective automation. Engineers working with physical systems -- manufacturing, propulsion, power generation, fabrication, etc. -- have always relied upon the fundamentals of control theory and feedback; for some reason we haven't included these ideas as part of the core of software engineering. Fix that. Now.

The book provides a very readable introduction to feedback control as applied to computer systems. I thought the introductory style was a refreshing change from the more traditional approach which tends to be very heavy on the mathematics. As an introductory text I think it serves it's purpose well. Being introductory means however there are a lot of things missing but combining this book with a more traditional control book would give readers the best of both worlds.

If you have a coding situation that smells even a little bit like a PID controller would help then you should get this book to provide good explanation of how dynamic control behaves in software. Good solid actionable info and clear presentation.Well done and thanks for filling a much needed segment of software fundamentals.

I'm only half way through, but it does compliment my Control Systems class (10 years back)...it is definitely helping me refresh for my current job.

If you write code for fun or for a livelihood, I recommend you check this book out.According to the book, Feedback Control is a topic well known to mechanical engineers, but not so much in the software industry. Feedback Control is about making smarter systems that can cope with dynamic environments. Many knobs that developers build into configuration can actually be automated with feedback loops.Examples given early in the book:* A Cache by tracking hit rate and changing the cache size* A Server Farm by tracking request latency and changing number of deployed server nodes* A Queueing System by tracking wait time and changing the number of workers* A Graphics Library by tracking memory consumption and changing the output resolutionThe book is well written. It starts out with practical examples and working code. It later introduces the deep theory and drops some math bombs. Don't worry, there is Python code for everything and you don't have to understand the math.It gives solid advice, like don't blindly use Feedback Control for optimization; optimization needs a higher level strategy guiding the process.Lastly, there are references for further reading, if you do want to work through more of the theory.The term Enterprise is thrown about, don't let this scare you away :) This is a valuable book for many types of software problems. A couple I've brainstormed while reading:* Controlling difficulty of a video game, to react to how skilled a player is* Controlling aspects of an animation* Controlling polling of APIs for fresh data* Driving load testing to find different scaling points (errors, high latency, etc)

I want to say one word... Just one word.,, Are you listening? * Feedback.*I couldn't resist using that line from the movie "The Graduate". Just as plastics became central to the industrial economy of the last century; feedback is becoming central to the real-time economy of this century. Effective use of information technology to enable companies to harness the power of feedback is the highest calling of the IT professional in this century and this economy. Companies that learn to harness feedback to guide their operations will be big winners in our real-time economy.Success in the real-time economy is about continuous response to change even more so than it is about being efficient. Efficiency is an Industrial age concept. Continuous response to change (while remaining efficient enough) is what we need to succeed in this age. That means feedback and understanding how to use feedback to guide and manage IT operations moment to moment.Mr. Janert presents the concepts of feedback in a clear and detailed manner. The IT professional who wants to specialize in the fine art and discipline of operating data centers in the real-time economy needs to understand feedback and how to apply it to run IT operations moment to moment as the real-time world unfolds. Janert presents specific examples of applying the principles of feedback to the tasks involved in running data centers. He illustrates his examples with samples of code and as you work through the code exercises, you get a clear understanding of how to apply these principles.The only reason I don't give this book a five star rating is that I wanted it to go beyond IT operations and show how feedback can also be applied to business operations. The same principles of feedback that govern the spin up of more virtual server instances in a cloud data center also apply to the order picking activity in a warehouse. The same principles of feedback that apply to the allocation of DRAM and storage space in a dynamic and unpredictable operations environment also apply to the allocation of money in a financial portfolio or the scheduling of delivery vehicles on a road network. IT pros need to appreciate this larger picture because it provides context to understand how IT and business operations can use feedback to work together in the real-time world.This book is for the devoted and highly skilled IT ops professional and for those who aspire to be highly skilled IT ops professionals. If I was a CIO or IT Ops Director running a data center of any size, I would strongly encourage all of my staff to read this book.

The following fact seems to be largely ignored: Feedback control is playing an increasing rôle for computer systems. Philipp K. Janert intends to explain to computer scientists feedback control, and especially PID (proportional-integral-derivative) controllers, i.e. the far most popular industrial feedback loop. The job is well done. It supersedes a book published ten years earlier by J.L. Hellerstein, Y. Diao, S. Parekh and D.M. Tilbury (Feedback Control of Computing Systems, Wiley, 2004). Let me nevertheless point out some weakness between the continuous-time setting in control theory and the often unclear discrete time for computer systems. This point is not enough detailed. I also do not understand why P.K. Janert, who states rightly that PIDs do not necessitate any precise mathematical modeling of the plant, devotes so much space to system identification. The new "model-free control" setting and the corresponding "intelligent" controllers (International Journal of Control, vol. 86, pp. 2228-2252, 2013) might lead to a more efficient approach, that is quite easier to implement. In spite of those shortcomings, which only reflect the necessity of further researches, I warmly recommend this book. It is today irreplaceable for anyone who wants to enter this fascinating engineering field.