MICROSERVICES ARCHITECTURE | SCALE CUBE | PART - 4
HTML-код
- Опубликовано: 5 мар 2020
- Learn strategies to scale the microservices.
#microservice #learnmicroservices #totorialssystemdesign #microservicestutorials
#systemdesigntips #systemdesign #computerscience #learnsystemdesign #interviewpreperation #amazoninterview #googleinterview #uberinterview #micrsoftinterview
One thing to be careful of when using data partitioning is that the expected load on each bucket is roughly equal. Consider partitioning by name in a country where there are only a few surnames: some buckets will get a huge amount of data and some none at all.
I believe what you call data partitioning in this video (splitting data from the same table)) is in fact called sharding. Partitioning is when a database is split into several databases - the concept essential to microservices that you touched upon in the previous video
Great video, thanks!
thanks so much, very informative and easy to understand videos
First like👍
Thank you for the cool video. All videos in the playlist except part 4 and 10 are private and not accessible. I just want to let you know if it is not by purpose
Hey Narendra,
Good video. But I do have some quick questions.
So when we scaled it further into S1(A-D), S2(E-H), S3(I-L), S4(M-P), S5(Q-T), S6(U-Z).
Do these scales also have their own DBs??
Because if it is, then isn't this Microservice is a BIG MESS?
And if it is, then why the heck these huge MNCs are still using it? How do they solve this issue?
Are all the microservices deployed on the same host(like using containers).? If yes then while horizontal scaling, will all the server machines have copies of all
the microservices?
I have a doubt, this individual services have to call REST APIs to interact with other service and other service will access to database , then overall performance doesnot get slow?
It depends on usecase, as when one service calls another service, It can call it async, and response from that service does not have to wait for response from another service.
Question: What are the advantages of using data partitioning over horizontal scaling ? To me, they both seem the same. One is just adding more nodes , the other is adding more nodes but handling specific requests per node
In horizontal scaling, the DB will remain common, and that can become a bottleneck. Hence, we can think of moving towards data partitioning to divide the DB, to serve only specific requests.
@@varunvats32 - Is it same as Data Sharding? Is data sharding same as what you explained or is it a different thing?
They are both different, You would use Horizontal scaling to a point but as your application grows and you understand the data pattern more, it could happen that traffic for A-N is more compared to N-Z partition so you would focus on scaling for that specific partition only rather then scale the whole service.
Beyond a point you would have to trade off and employ other strategies like service do-composition.
@@NeerajPahuja Data partition would apply to data this applies to service
Microservices is a huge mess.