Spark Architecture in 3 minutes| Spark components | How spark works
ฝัง
- เผยแพร่เมื่อ 10 เม.ย. 2021
- Spark is one of the most prominent and widely used processing framework in Bigdata world. This videos explains the core components and architecture of spark with a real world example in just 3 minutes.
- วิทยาศาสตร์และเทคโนโลยี
Finally it got cleared to me after reading here and there . thank you .
Ma'am you taught amazing 😍😍 very less time consuming lecture but perfect... Keep it up
I havent seen such a lucid way of explaining something this complex concept. Great work!
Thanks Bitthal
Excellent way of explaining things in a most simplified manner.
Looking forward to more videos on Spark.
Thanks Satish
very simple and nice explanation. Thank you for posting this video
Example was very good for beginners
This is the only video on TH-cam which clarified my doubts. Thanks!!
Thanks showbhik
Thanks showbhik
Great Video, I have learned alot.
Thank you
finally, a video that simplifies spark, amazing, keep the videos coming please!!
Thanks
Simple example and easy way of explaining an important concept.. thanks!
Thanks seemanthini
This video is better than 1 hour course on spark . Thanks
Thanks puneet
This made far easy to understand Spark Architecture. Thank u ma'am, you are great
Thanks Ashish
Thank you ma'am..👍
Lot of information presented in simple way for everyone to understand. 👍
Thanks sriganesh
Very nice video and it covers everything related to the Spark architecture in just 5 minutes. Keep sharing new videos.
Thanks sandeep
Great video
Very Informative, with full of clarity, Thank you.
thanks
Vividly explained. Thanks mam
Thanks saurabh
Nice explanation.. Pls keep videos 🎥 like this
Very clearly explained, really appreciate all your efforts
Thanks Dhruv
way of explanation is .....just amazing
Thanks
Thank you so.................. much for this
Just woow, very simple explanation of a complex cluster overview..
Thanks.
Thanks
Amazing!
Thank you mam for excellent way of teaching Spark.
Thanks
Simple and easy to understand. Thanks. I like doodle way of explaining concepts :)
Thanks sheik
Grate nice explanation
just love the way u explained it ;) Really appreciated mam.
Thanks minakshee
I wish I could give 1000 likes. You’re an excellent teacher!
Thanks
the best explanation ever great work !
Thanks
one of the bests
Thanks vikas
Excellent example 👏
Thanks
Beautifully explained short video 👏👏
thanks lakshmikanth
Amazing content ! Really appreciate the understanding and approach to explain...looking fwd to more
Thanks Aditya
Extraordinary mam
Thanks harika
It's the best video I've seen so far on spark architecture..awesome..keep going..
Thanks
Your videos really help me in clearing my interviews and getting the job, now I need some support for my new job. Can you either post videos regarding how pyspark class object work in backend, how parquet/csv reading writing work in distributed environment like which data read by each executor, how to do pagination of some order data etc.
If possible provide your LinkedIn profile url or suggest some way to connect
VERY HELPFUL BEST EXPLANATION
Thanks
Thank you for this video.
Thanks
Superb delivery
Thanks Hemanth
Extremely good explanation
Thanks mou
Very helpful, it would be great if you can take an example and illustrate how the data chunking happens
Thank you for such a goooood explanation :D
thanks sheereen
Nice explanation. For 1 GB input data for a batch processing how can we decide how can we decide the cluster size, no. of nodes or no. of executors ? Could you please explain Thanks ma’am
Saw this video.. content looks promising... great job
Thanks
Very Nice Explanation
Thanks Naresh
Great work ma'am
Thanks Shivratan
Thanks a lot mam
Thanks Sandy
Subscribed ✅
Very nice explanation,
Thanks Vidya
Amazing explanation mam 😊😊👍
Thanks ravi
Great explanation Ma'am, please add more videos and arrange it in seq. under playlist
Thanks Chetan. yes there are more videos coming up. stay tuned
good explanation...
Thanks
Excelent video :)!
Thanks
This was marked to know that I'm here on 3/10/2023
How did you make such a good visual explanation? Which tool you used to draw sketches ? Pls guide 🙏
Very nice 👍
Thanks
@@BigDataThoughts did you explained in this style big query also. ? improvement in this video can be summary in slow way. please dont get hurt because i gave comment. you did really well in video. excellent explanation.
Great video!
I have one question though. Is it my correct understanding that each student which got the coin bag is same as how data is partitioned.
i.e. 1 student = 1 data partition?
1 data partition is operated on by 1 slot/task
One executor in one core and 2 partitions are assigned so one by one will execute. My que is if 10 tasks are there then these tasks wil execute parallelly or sequentially in partition level
A task operates on a partition of data. Tasks do run in parallel. If you have multiple cores you can specify how many cores will a executor use. The number of concurrent tasks an executor can run is equal to the cores assigned..
Really informative Shreya. One quick question. Stage will run sequentially depending on use case and Tasks will run in parallel?
Thanks Madhu. Stage may run sequentially or in parallel depending upon whether they have dependency or not. Typically a stage will have multiple tasks running in parallel on a different set of data but doing the same set of operations that the stage contains.
Nice video. Really liked it. So you said one node can act as driver. I want to know what is the best practise here? I usually submit jobs by doing SSH to master node (atleast in GCP dataproc) and then submit job. So should I consider my master node as driver? Is it right to do that way?
To give an example if we are using Yarn - when you are submitting a spark job in cluster mode. Container where the Application Master runs acts as Master node (driver) and the containers where all the executor process runs the tasks are called Slave Node. When the job gets submitted first the spark submit calls resource manager which in turn starts the application master and from there driver takes over.
Mam, I have one question.
If spark has to write a data to sql database and as our data is broken on to multiple worker nodes, so is it driver who establishes single connection with sql db or it is worker nodes who establishes multiple parallel connections ?
When Spark writes data to a SQL database, it is the driver program that establishes a connection with the database and manages the write process. Each worker node will write its portion of the data to the database through this single connection established by the driver
@@BigDataThoughts Thank You Mam for the elaborated explanation.
❤
could you pls explain more on parttiton
The Dataset is divided into partitions and each partition is the unit on which a task works. That's the input to task.
Can you help in understanding RDD
RDD are resilient distributed datasets and they are the lowest abstraction in spark. check this video - th-cam.com/video/DuWGBMF7ARc/w-d-xo.html
At the start of the video i was so happy seing all the diagrams..
Later got fully confused & felt complicated and i didnt understand well 😢
Can you please turn on the subtitles? thank you
Can you make it slow to follow. I felt this was fast to get to know the terms.
Good playlist for Spark
th-cam.com/play/PL1RS9FR9qIPEAtSWX3rKLVcRWoaBDqVBV.html
Thanks
Great Video....!! Appreciate your efforts!🎉
One question, Where does a cluster manager fit in in this architecture? What role does it play in comparison with driver?
Cluster manager's job is to provide resources for job execution. Ex - yarn, mesos etc. Driver is the one controlling the overall job execution and which executors take part in the job
@@BigDataThoughts ohh okay!! Thank you!!
18/april/2024