With containerization gaining momentum and kubernetes promising better deployment models, a question first rises; what kind of a model should one support when it comes to large scale streaming type of applications?
High volume use cases are understood by all and many of them have talked, others have been able to deploy high volume, huge scalable architecture model.
Dockerization or containerization is much needed within the healthcare and financial domains for sure, not to mention manufacturing and retail. However, feeble and shabby manual installations prevail within such organizations.
It is some times sad, pitiable and annoying to see such haphazard and immature deployments. People who have deployed such models would have been better off without such deployments and would have invested on learning about containerization. Nothing but words and BS flow from supposedly architecture discussions while consultants, engineers and vendor companies struggle through, spending half the time battling compliance and policiy issues and above all the jurisdiction problems such as who will do what and so forth.
Reminds me of a quote :-
“People who think they know everything are a great annoyance to those of us who do” ~ Isaac Asimov.
Its ok not to know; but it is completely stupid to oppose a budding model and kill it instantly.
As technology shifts and evolve into something beautiful, it is important for people to embrace aggressive technical discussions and forget emotional disturbances that can interfere with personal agendas and issues. Let there be emotional intelligence rather than personal emotional turbulences. Not many come to the enterprise to marry and please hey, nobody is inside to fight with people either.
By writing off, saying, such stupidity will exist within this world has now come up to a point where one needs to really “think differently“, isn’t it time to “think differently” ?
A search on the internet on scalable models or large scale deployment give architecture descriptions and youtune like places spending time on details of containerization internals and so forth. But not much exist in the form or diagrams or architecture model that can be adopted to simply use it. Individuals with eagerness and enthusiasms have been shot at during these discussions and have lost their flames. Let me stop before the subject diverts.
Below see if the model makes sense. Please feel free to give your opinions. If you have questions, do write it here. Collaborate for the sake of knowledge THEN COLLABORATE FOR THE SAKE OF MONEY.
******Architecture diagram below*****CLICK TO VIEW AS LARGE IMAGE*******
A few points to note.
1. This is a single unite within a cluster( I don’t want to call it worker because there are multiple workers within multiple containers here)
2. Number of boxes DO NOT represent number of components such as HA PROXY or NGINX. Containerization takes Docker into perspectives. Changing the box to COREOS may not be that of an efficient model.
3. Number of nodes is provided as an approximation (125 nodes)
4. Within a SWARM environment, nodes can be reduced. This model depicts a single unit cluster. (Worker in generic term, however I do not want to call it a WORKER Node)
5. This is to be viewed from an Application standpoint and not from INFRASTRUCTURE standpoint.
Leave your comments, whatever it may be. It is from the many mistakes that one learns a lot. If you ask, I will reply.