Data Consistency | Strong Consistency vs. Eventual Consistency | System Design for Beginners

[@joflve]

The only downside of your channel is that I cannot get new content every minute. The most clear explanations I've seen for technical interview topics.

[@harshmohan8419]

Same hehe. I need good system design soon. This channel only way left and way she explain is legends.

[@ShiranAfergan]

Thanks! Made me laugh and blush 😂😂

[@adeolaoladeji1167]

This is the most beginner friendly channel that treats system design. Thanks for clarifying all the ambiguous concepts.

[@stefanadranca560]

Love the visuals and how clear your explanations are!! Thank you!❤

[@andrewkicha1628]

Amazing job done explaining the concept, I really liked the visualizations and the structure of the information

[@ordered_saddle5]

thank you for making it simple, as it is

[@naweedjaulim6058]

Best System Design channel I have come across so far with the illustrations! Thank you.

[@ShiranAfergan]

Thank you! 😊

[@gcheese25]

loved your explanation so much! wish me good luck for my system design interview

[@parthabhowmik4747]

You are so good at explaining things, thanks.

[@kirillzlobin7135]

This channel has a lot of useful information. Thank you for sharing

[@brajagopalmukherjee1588]

Mam I am following systems design primer GitHub one and your video along side to prepare my system design video

[@ankithap5219]

Very clear and consise explanations. Thank you!

[@Canda-fh4xc]

Easy and clear explanation. Thank you

[@ganapathinaik5446]

Nice explanation and very informative Thank you. Waiting for videos on database partitioning and sharding.

[@ShiranAfergan]

Thanks! database videos are definitely coming :)

[@benparker8000]

Glad I discovered your channel. I'm brushing up on these concepts before an interview and these videos are very helpful. I checked out Guarav Sen as recommended in another video as well and he's excellent. Your advice has made this learning process much easier. Thanks!

[@yuanliu5945]

The best video explaining system design!!

[@ShiranAfergan]

Thank you! 😊

[@peternicola6429]

you earned a new subscriber, thank you for your help 🙌

[@caiobittencourt9595]

So good, very easy to understand!

[@minhhuynhthe]

thanks for your video. It helped me a lot.

[@khanriza]

Great video

[@shubhamladha3346]

Awesome

[@harshmohan8419]

Loving Yaa. Need more awesome knowledge.. Soon. Take care

[@ShiranAfergan]

☺️ More is coming :)

[@harshmohan8419]

@@ShiranAfergan eagerly waiting..........

[@MrEG12341]

Thanks Shiran. When using a cloud provider (as aws), what are the pros and cons of using the managed read replica?

[@ShiranAfergan]

Like any other managed service, it saves you the trouble of doing some things manually. You won’t have to burden yourself with setup, upgrades, backup, and more. The downside of managed deployments vs unmanaged is less control over configuration. There are also other limitations (storage, number of instances…) and it costs more.

[@saiashok28]

You deserve more subscribers Shiran!!!❤️

[@gurpreetvirdi6967]

Awesomely explained. Please keep making more content :) . Cheers!!

[@neiljohari6815]

Thanks! QQ: do systems actually block read requests, or is it more reasonable to block the write’s acknowledgement until a synchronous replication quorum is reached? I’m not sure how you’d be able to block reads fast enough on geographically separate DCs (since you’d need to send a message which will still suffer from network latency)

[@ak-ot2wn]

5:21 - what if all 3 clients are updating the view_count value on each of these nodes? Will the view_count value become 21 instead of 23 or somehow magically it will become 23 on all 3 machines?

[@nikhilpatil9654]

great explanation

[@GidzPaul]

Thank you!

[@prudhvisaisuggala7677]

Nice video

[@SPribyt]

great!

[@AjitKumar-wm5cl]

Where are you now a days? No more video for many months. Please make some technical video and share with us. Liked your videos

[@legenddesigning7111]

keep up

[@ajisegirisunday2885]

Nice video but I have a question. Let's I have my db operator deployed on kubernetes as statefulset and I set replicas to two or more.. How does kubernetes managed the replication.. Is it strong consistency or eventual? Can I configure it to my requirements?

[@ShiranAfergan]

Thanks :) Kubernetes is not related to this. The consistency of a database depends on the database you choose. For example Postgres provides strong consistency while Cassandra is more eventually consistent (might be configurable)

[@ajisegirisunday2885]

@@ShiranAfergan thanks i know better now

[@keistzenon9593]

Is that the reason why comments on youtube videos sometimes seem to disappear or appear much later? (besides some filtering)

[@yonishachar1887]

את דוברת אנגלית מלידה? יש לך תוכן ברמה גבוהה שמועבר באופן פשוט וברור. ממש אהבתי!

[@ShiranAfergan]

לא, לא מלידה. הרבה טלוויזיה וסרטים מגיל קטן עזרו 😆 תודה על התגובה! שמחה שאהבת :)

[@atulmishra1517]

In case of strong consistency, how are we going to block reads while replicating data?

[@redhot678]

A question, what would happend if the time until the replication message is too long and we get a get views request in the middle

[@dmitriybaram4595]

Shiran, can you elaborate on how exactly requests to replicated servers can be blocked if clients are not aware of the replication process that happens asynchronously?

[@brlvldvlsmrtnz]

The way I understand it, it's actually the write that blocks until all the replicas have also performed the write

[@arneishprateek6444]

Your read request would need to be sent to multiple servers and at least "R" of them should agree on the value before it's returned to you. Similarly, your write request would need to be completed on "W" servers before it's returned as completed. If R+W>N where N is the total number of servers, your system is strongly consistent since then there is at least one server that must contain the latest value. This, of course, comes at the cost of performance as she mentioned in the video - since you now need to wait on R (or W) servers to return. The lower the values of R and/or W, the lower the consistency of your system but higher the availability (and performance). In the extreme case, R=1 and W=1 would imply the most available but least consistent system. (Source: System Design by Alex Xu)

[@yogeshdharya3857]

considering the comment section , seems ur videos are being recommended in India more ! Who knows if eventual consistency is to be blamed for this 🤣! I like those plants waving cause of the wind !

[@ShiranAfergan]

Haha yes maybe :)

[@1roadrage1]

Comment of substance. trust me bro.