WEBVTT

00:04.300 --> 00:07.250
We have just defined what is a Kafka.

00:07.270 --> 00:14.380
It is distributed by subscriber messaging system and in every published subscribe system, messages

00:14.470 --> 00:16.260
should be stored somewhere.

00:16.660 --> 00:25.120
And also publishers should be able to send to where those messages and the subscribers should be able

00:25.120 --> 00:27.430
to read from somewhere those messages.

00:27.580 --> 00:32.440
And in a Kafka, brokers are responsible for all those operations.

00:32.620 --> 00:38.170
They store messages and they sell publishers and subscribers.

00:38.500 --> 00:42.100
But publishers in a Kafka are gold producers.

00:42.100 --> 00:49.000
They produce messages to Kafka brokers and subscribers in a picture, Kafka A consumers.

00:49.180 --> 00:52.210
They consume messages from Kafka brokers.

00:52.390 --> 01:00.010
And most interesting that at every single physical server of Utah, Sarah, you may run multiple Kafka

01:00.010 --> 01:05.760
broadcasts and they will be completely independent and you are able to create even Kafka Gloucester

01:05.770 --> 01:07.220
on a single computer.

01:07.870 --> 01:14.800
This Kafka broadcast, again, its main responsibility is up following first receive messages from producers.

01:15.250 --> 01:22.510
Second thought those messages and thought give ability for consumers to read those messages, that's

01:22.510 --> 01:22.810
all.

01:23.260 --> 01:28.360
And the Kafka broker simply stores messages in files on a hard drive.

01:28.600 --> 01:35.560
And producers are able to append messages to those files and consumers are able to read from those files

01:35.680 --> 01:36.780
that that simple.

01:37.510 --> 01:44.830
Also, it is possible to have multiple producers and multiple consumers like on this diagram, and multiple

01:44.830 --> 01:51.580
producers are able to simultaneously produce messages to Kafka broker and multiple consumers are able

01:51.580 --> 01:55.450
to simultaneously read messages from Kafka Breuker.

01:55.900 --> 02:02.890
And of course, messages may be produced and consumed asynchronously on different moments of time and

02:02.890 --> 02:06.430
noticed that there was a single weak point on this diagram.

02:06.580 --> 02:07.660
And it is Breuker.

02:08.020 --> 02:14.470
If this Breuker fails, nobody will be able to sell producers and consumers this way.

02:14.560 --> 02:23.050
Usually nobody runs a project like that with just a single broker and instead broker Gloster is created.

02:23.320 --> 02:24.730
Let's discuss what is broken.

02:24.730 --> 02:26.830
Gloster in the next lecture by.