Kubernetes Architecture in 7 minutes | K8s explained — Transcript

Full transcript

1. 0:00in this video I will explain you
2. 0:01kubernetes architecture in simplest
3. 0:03manner possible we go deep down in the
4. 0:06kubernetes cluster understanding
5. 0:07different components and how they
6. 0:09actually work so make sure you watch
7. 0:10this video till the end let's start so
8. 0:13we start with creating a cluster you can
9. 0:15create a kubernetes cluster locally
10. 0:17either using tools like mini Cube or
11. 0:19cube ADM optionally you can also use
12. 0:21probably cloud provider services like
13. 0:23eks and AWS AKs in aure gka in Google
14. 0:26Cloud to create kubernetes clusters
15. 0:29inside this kubernetes this cluster we
16. 0:30have two different types of machines or
17. 0:32as we call them nodes so we have a
18. 0:35master node which is responsible to
19. 0:37handled everything inside the cluster
20. 0:40along with Master node we also have
21. 0:42another kind of node known as worker
22. 0:44node this worker node is actually
23. 0:47responsible to run your application
24. 0:49containers inside kubernetes cluster so
25. 0:52you have Master node that is when to
26. 0:54handle everything inside the cluster and
27. 0:56you have worker notes which are
28. 0:57responsible to run your application
29. 0:59containers inside the kubernetes cluster
30. 1:02let's start with worker nodes and look
31. 1:03at different components inside it as we
32. 1:06all know worker nodes are responsible to
33. 1:08run application containers and they run
34. 1:10this application containers inside
35. 1:11something known as pods pods are
36. 1:14smallest Deployable units in the
37. 1:15kubernetes tester and the Pod can hold
38. 1:18one or more containers a single worker
39. 1:20node can have more than one pods or it
40. 1:22can have a single pod as well so these
41. 1:25containers are actually running inside
42. 1:28object kuet is object known as pods to
43. 1:31deploy and manage pods inside this
44. 1:33worker node we have an agent named as
45. 1:35cuet this cuet agent is running on all
46. 1:38worker nodes and it gets information
47. 1:40from the master node to decide where
48. 1:42should be the P running and how to
49. 1:44manage it along with this for a worker
50. 1:46machine to create and manage containers
51. 1:48we need to have a container runtime
52. 1:50interface or a software like Docker so
53. 1:52worker machine will also contain CRI or
54. 1:55container Endtime interface such as
55. 1:57Docker here other options are container
56. 1:59d locket Etc which is supported by cetes
57. 2:02as well also to enable communication
58. 2:05between these two ports or to make these
59. 2:07pods or application accessible to the Ed
60. 2:09users we have another networking
61. 2:11component in the worker node which is
62. 2:13known as cube proxy qoxy is used for
63. 2:17Port toport communication also for load
64. 2:19balancing service proxy and to make your
65. 2:21application accessible to the end users
66. 2:24so these are different components inside
67. 2:26the worker nodes you have q proxy for
68. 2:28port-to-port communication and expose
69. 2:30your application you have container
70. 2:31runtime interface to create and delete
71. 2:33containers you have cuet an agent that
72. 2:36is running on every worker node which
73. 2:38get information from the master node and
74. 2:40handle the workloads inside the worker
75. 2:42nodes along with this you can have more
76. 2:44add-ons or plugins that you install on
77. 2:46your worker notes but these are all the
78. 2:48important components inside your worker
79. 2:50notes now that we have understood the
80. 2:52different components inside worker nodes
81. 2:54and how they actually work let's look at
82. 2:56the different components inside Master
83. 2:58node so so Master node or control plane
84. 3:01has four very important components that
85. 3:03are responsible to manage everything
86. 3:05inside the cluster the first component
87. 3:07is the cube API server this Cube API
88. 3:10server is like an entry point for
89. 3:12everything you do inside the cluster so
90. 3:14let's say you are a devops engineer and
91. 3:15you want to create a new pod or maybe
92. 3:17delete a pod or just get the status if
93. 3:19you want to do anything inside cluster
94. 3:20you will have to talk to the cube API
95. 3:23server so you can talk to cube API
96. 3:25server either by running commands using
97. 3:27Cube CTL command line tool either using
98. 3:29Cube uh Cube it is dashboard or maybe
99. 3:31using sdks or anything so you will talk
100. 3:34to the cube API server Cube API server
101. 3:36will validate if the request is okay if
102. 3:38the user is authenticated or not and if
103. 3:41it works then it will get the
104. 3:42information from the other components
105. 3:44and provide it to the end user or to the
106. 3:46dev
107. 3:48optionor so API server is like an entry
108. 3:51point which validates the request and
109. 3:52also authenticates the user next Master
110. 3:55node component is the scheduler
111. 3:57scheduler is used to assign pods to the
112. 4:00worker nodes let's say you made a
113. 4:01request to the API server to create a
114. 4:03new pod scheder is going to check which
115. 4:06nodes have enough capacity to launch
116. 4:08this new
117. 4:09pod once the nodes on which the pods are
118. 4:11going to be run is decided by the
119. 4:13scheduler scheduler provides this
120. 4:14information to the cube API server Cube
121. 4:16API server gives this information to the
122. 4:19cubet running on that particular node
123. 4:22and thus it creates the new pod on the
124. 4:25worker note so scheder is used to assign
125. 4:27pods on the notes based on resource
126. 4:29capacity all the Manifest files that you
127. 4:31have defined for your
128. 4:33cluster now what if there's an issue
129. 4:35inside the cluster or one of these ports
130. 4:37shows an error or gets deleted that's
131. 4:40when the next kubernetes Master node
132. 4:41components comes into play so next
133. 4:43kubernetes component is the controller
134. 4:45manager controller manager is used to
135. 4:47monitor the entire cluster and it makes
136. 4:49sures that everything is as you have
137. 4:51defined inside your manifested files so
138. 4:54let's say you have said you want to have
139. 4:55four pods running it will make sure that
140. 4:57the four pods are running all the times
141. 4:59if that's not the case it'll tell API
142. 5:01server which will recreate the P
143. 5:03following the same procedure with the
144. 5:04scheduler first decides what nodes to
145. 5:06create the pods on then gives the
146. 5:08information to API server and then cuade
147. 5:10creates it so controller manager is used
148. 5:13to Monitor and make sure everything is
149. 5:14working as you have defined and it does
150. 5:16that by comparing the desired state with
151. 5:19the actual state of the cluster but
152. 5:21where is this cluster information or
153. 5:23cluster data stored it is stored inside
154. 5:25the new it is stored inside the next
155. 5:27Master node component which is hcd hcd
156. 5:30is a key value store that stores all the
157. 5:32information about the cluster what
158. 5:34application is running what pods are
159. 5:36running on which nodes and everything
160. 5:37else one thing to note is hcd can only
161. 5:40talk to the cube API server so if you
162. 5:42want to get any information from the hcd
163. 5:44it has to be through the cube API server
164. 5:47so you can see Cube API server is like a
165. 5:49center of communication for every
166. 5:51component inside the master node as well
167. 5:53as to the CET which is running on the
168. 5:55worker node so this is the kubernetes
169. 5:58architecture we have Master node and the
170. 6:01worker node Master node is responsible
171. 6:03for handling workloads inside the
172. 6:04cluster worker nodes are actually
173. 6:06running the application workloads here
174. 6:08inside worker node we have the parts
175. 6:10which is the smallest unit responsible
176. 6:12to run the application container we have
177. 6:14cuate and agent which is going to run on
178. 6:16all the worker nodes and also
179. 6:18responsible to create pods next we have
180. 6:20con the runtime interface and we also
181. 6:22have q proxy for networking and for
182. 6:24exposing our application inside Master
183. 6:26node we have QBE API server which is the
184. 6:28entry point and the authentic ation
185. 6:29system for every request that you make
186. 6:31it also exposes a kubernetes API for the
187. 6:34end user to start using it next
188. 6:36component is a scheder which is used to
189. 6:37assign ports to the notes then we have
190. 6:40controller manager that makes sure
191. 6:42everything is working as you want or as
192. 6:44you have desired and all the cluster
193. 6:46information is stored inside hcd so so
194. 6:48this is a simplified explanation on ku's
195. 6:50architecture I hope this video was
196. 6:52informative if you have any questions
197. 6:54any doubt do let me know in the comment
198. 6:55section and for more information I would
199. 6:57recommend checking out my another video
200. 6:59which explains what is kubernetes and
201. 7:01how it actually works do check it out
202. 7:03thank you and have a good day