Essentials Ks during my Kovid Year
“ Koan, Kafka, Karl Pooper, Kats, Kubernetes, Kloud”
Entailing my nerd journey how my year of Covid went actually from March 2020 — Feb 2021. It might go longer, but for now, the proof of pudding is in this writing
1. Koan
The concept of Koan is “unanswered question” or a riddle which nobody can answer, but every attempt at answering reflects about the individual who attempts, and everytime it throws in a perspective otherwise. In Zen Buddhism, Koan is meant to break boundaries and create an open mind. In everyday nerding, ESSENTIAL to have that open mind
Why Koan? Because there is tendency sometime to assume everything as objective real and settled in stone, when it is not. Some burning question may never get answered. The key is then to act & do what ? The absolutism limits our capacity to think, forget even help in absorbing a counter perspective. Constantly knowing that there is something you dont know and you are merely interpreting as it “occurss” to you
Quoting from book “Three Laws of Performance” : So what exactly does ‘occur’ mean? We mean something beyond perception and subjective experience. We mean the reality that arises within and from your perspectives on the situation. Again we are not saying that there isn’t a “real world”. We are merely pointing out that our actions relate to how the world occurs to us, not the way that it actually is.
2. Kafka
While building software systmes for small and large business, the capability to act fast in response to any market or customer activity takes the central stage. Message driven architecture of software systems, are enablers of such business agility or any enterprise looking for some “digital transformation” for themselves. Ticketing, Fraud detection, Reminders, and other event based automation, you have to get Kafka in your architecture or its equivalents (Spark, PubSub, Kinesis). This is Not rigid service definitions like ESB(Enterprise Service Bus), but more flexible and optimal when applied with mindset of Domain Driven Design (DDD). Again, I had a decent stint with Kafka in Covid year and I consider it ESSENTIAL in post-Covid world as well for any enterprise looking for agility. Check for yourself otherwise reading here
3. Karl Popper
Investor George Soros is self-claimed follower of professor Karl Popper. Karl is behind the theory of “falsification”. His book ‘The Logic of Scientification’ is a must read and without giving any concrete example of modern times, it gives you sufficient insights how quoting science and politcization of science and funding selective themed researches can lead any society in to believing new prophets who stand behind any falsifiable research.
One of Popper’s great insights was that we should fear those who claim to have a monopoly on truth and exclusive knowledge of the One True Way, particularly if they hold power. Popper also finds good mention in this Stanford article
Almost ESSENTIAL in post-truth world I must say
4. Kats
Well, where to begin, getting one cat itself was an experience, and then I got myself 3 cats. Then came the useful learnings of dealing with psychology, insecurities, and irrational cringes to deal with, and most important lessson while getting more than one cat was — how to bond them. Failed twice and then learned 10 day long process of keeping two cats separate and running the sponge over their whiskers and exchanging those smells till their hissing stops
Important Lesson: The team formation lectures of “form, norm, storm” are over-rated while building spatial & diverse teams. I would say (1) Bring different diverse people together and separate them with their tasks and deliverables while in the same team (2) Keep them all aware of each other’s thinking process & working style every day on how they perceive a problem and how they want to approach solving it,WITHOUT others influence (3) Let them finish their individual tasks, and then put them in team together to calibrate themselve to task up
My Kats context example does not finish alone with how petting cats can give you valuable insights on team work as leader or manager, but same goes for business collaborations as well
Back in 1991, Geoffery Moore in his famous book “Crossing the Chasm” wrote the following phenomenon how large businesses seek younger business products to keep extending its own life time ( Yeah those were not the days we talked about Corpoorate-Startup innovation partnership, or blood transfusion technically), and such advertisements were rolled out in newspapers in late 1980s
“Large, well-heeled company with established distribution channels and aging product line seeks small, entrepreneurial, cash-starved technology leader with hot new product. Photos available upon request. Write box no….” (Page 92, Partner and Allies, in the book “Crossing the Chasm” )
What went wrong often in such cases? The words mentioned are ‘company culture’ , ‘each side has probably misrepresented itself one way or another during negotiations’, ‘Decision cycles are widely out of sync with each other’ in the paragraph that follows. Now just think back of those individuals as Kats, and then what early signs to collaorate and calibrate you would watch out for and set milestones to ensure the partnership prevails ?
5. Kubernetes
The ESSENTIAL container orchestration framework to distribute your software deployment to scale over different networks, originally open sourced by Google and adopted by CNCF (Cloud Native Computing Foundation) is the new default and defacto way of deploying your software in scalable manner. With advent of more cloud and distributed computing, this is already setting a foot print in every technology stack … except Microsoft which might want to bring in an alternative when it becomes able to fully utilize Linux instead of PowerShell. But for now even Microsoft Azure has to have support for K8s (written shortly for Kubernetes ) to gain its own acceptance as cloud vendor
There is bit sociological lesson also in distributed computing as all nodes are not created equal, the concept of “Taint” and “Toleration” is gold in K8s as it provides the flexible means to work with nodes and pods with inequal capability and capacity like absolutely real world, while most of distributed computing lessons while being taught axiomatically tend to forget the inherent inequal nature of computing resources, and the dynamism they constitute in any software application deployed in any cluster of such inequal resources. Equal is unfair anyone? Because deployment orchestration, replica maintenance, decide to do a daemonset or service, all keep the realistic context of inequality of computing resources and serving dynamic nature of incoming user requests
6. Kloud
A common myth that everyone carries is shipping your software deployment over to Kloud is just like a server migration or software installation. Security concerns aside (where most of the time currently goes discussing), the main tangent is those ‘mostly’ who want to deploy their application on cloud do not know precisely how much CPU, memory space their software application needs. A series of horrendous decisions is certain, and some might trigger refactoring, breaking up into microservices, code cleanup, rigorous work with K8s in-premises before those softwares and application logic can be optimally thought of deployed over Kloud
And FaaS/Serverless is yet to take over, personally I already find Lambda and Kinesis kinda infrastructure in cloud is still evolving but also pricey. Once that cost comes crashing down and infrastructure components mature, those who currently seek cloud migrations and adapting/refactoring their softwares for that, will be forced to run into another spiral of breaking-up the micro-services into micro chunks of functions
And guess what all it would be called then when a legacy application has to move to cloud ? Megacloud ?? if migrated without refactoring :)
Furthermore, recycling of infrastructure on cloud serverless platforms is known as the freeze/thaw cycle ( Yes that is the climate change your deployed application has to face in the Kloud ). On AWS Lambda, after approx 45 mins of inactivity, subsequent calls to an endpoint reveal no trace of original function containers or their host VMs. Consequently, future calls force initialization of new server infrastructure adding latency to service response times because of inherent design of cloud infrastructure (Suggested reading here). So if you are migrating a monolithic continuous transactions processing system with certain low activity phases, the application predicaments are all going to assessed afresh, and how the optimal resource allocation or reservations are going to be orchestrated. Kloud is ESSENTIAL and almost everyone has just begun to grasp and grapple with deployed functions representing an enterprise
Thats all for now, the year from March2020 — Feb 2021 end has been a full circle of journey. I could fathom these Ks, while next year could bring more surprises with insights learnings and procreations!
