doubt, so now 1st leaf node (block) contains 3 rows, when we update its 1st row to an extent that it occupies whole space of the node leaf (block), will the rebalancing happen on the 1st leaf node & a new node will be created for 2nd & 3rd row ? how will this look like ?

1. Are indexes sorting again & again on inserting new data. 2. then does data also move from 1 block to other block after index resorting.

I love the fact that you make reference to real life applications and examples

but for the file case also we can't we do a similar operation of reading the entire file, bringing it into memory and then shifting the data around and finally flushing it back to disk. What advantage does B+ tree offer, I mean the only one which I can think if is that in case of B+ tress you wont have to touch the entire data but rather only those specific 100 rows, then can't have a design where we store 100 rows on file and the next 100 on another file and achieve the same output/performance ?

You mentioned in the first approach that while updating a row we cannot automatically move beyond the the size of the width of the row but later as you mentioned we have a fixed row size so in what case we will be exceeding the width and overwriting other data for the naive file storage approach?

Thank you! You are very helpful 🙏

What iPad note application are you using?

great video, a fan from linkedin

Hi Arpit, please do correct me if I'm wrong. Load factor signifies how much the hashtable is filled. So mathematically it should be (# of Entries / Size of Hashtable) So if we say that Load factor = 50%, it should be = 1/ 2 => 0.5 But in this video you have mentioned for 100 / 75 for a LF = 75%. Shouldn't it be the other way ( 75 / 100 ) ?

Great video. I have fxcxew questions: 18:30, Would leader node applies the updates to the table in realtime or async updates like followers in case of eventual consistent reads. Also, what happens to writes in consistent writes? Who updates partition data in Metadata service? What will happen if my data storage node has capacity of 1000IOPS and I have two partitions P1, P2 of 400 each of two different tables T1, T2(of different customers). Now client requests for 800 IOPS on T1 for P1. Now the required IOPS are 800(T1, P1)+400(T2, P2)=1200IOPS but server only supports 1000. How the server would handle this load in case of adaptive load? Also, please consider the case when partition 2 also needs at-least 400IOPS at any time. The customer don't want any latency as I've paid.

Great video, Arpit. QQ: Isn't scatter gather considered to be bad practice in distributed env? What if you scatter a query to one of the leaf node and that takes longer time to return the response? It will lead to slowness of entire scatter gather process right?

Next Cockroach DB please

How the read will work if copy will take days, please explain this part too

Superb explanation Arpit with good examples helps to grasp the concepts easily.

SELECT u.user_id, u.username, COUNT(b.blog_id) AS blog_count FROM users u LEFT JOIN blogs b ON u.user_id = b.user_id GROUP BY u.user_id, u.username ORDER BY blog_count DESC;

what about using pessismistic locking + skip locked? this will help in threads not getting blocked right?

Can I do these 2 projects in C++ ? Will it help to learn the same or have to go with C ?

Wow explanation Arpit

Thanks Arpit. These explanantions makes learning fun.

thanks a lot , this concept made my day

are you gujarati bhaiya ?

Thanks a lot for this video, would really love to understand working of @Timer and the nuances of implementing the same. Can you explain that ?

Loved the hack to make chaining cache friendly. Would also please provide the links to papers you have read?

One of the best tech podcast which talk real to the ground. sidhi baat no bakwas. Many learning hidden for new startup founder and CTOs.

@arpit bhayani - Can you share slides or notes.

If you have gone through the video: In the final approach, incrementing the atomic operations are still a bottleneck. They will be causing thread contention. ( Multiple threads waiting to do the atomic operation ) Fixing this will be an interesting challenge. Kudos to the maker, this is an amazing video.

This is exactly my understanding of a protocol is, after working on several protocols for the last few years! It’s amazing to see such content for free! This drawn me the interest to check your other videos too! Keep up the good work brother!😇

Mine actually took 35.2x seconds, very fair

17:04 hard deleting is hard on the DB 😂 So rate limiting can just be part of business logic without explicitly specifying it? aka not necessarily a separate component or something complex

usefull content...thanks!

Hey Arpit, Can you please make a video about Aws aurora ?

how does joins work with vertical sharding where individual tables are moved to different databases, complex queries will be affected right?

Quick question : How can I keep fetching live data from an API endpoint and ingesting it into a database . Right now , I use a cron job to fetch via a script the newly updated records once a day . How can this be done in realtime ?

Thanks for the great video. Follow up question: How is "scatter gather" fast? What if one of the nodes to which query was sent took longer than others? In that case it would delay the entire process, isn't it?

I found this video and your entire channel both interesting and amusing :) Fantastic videos. Simple to understand yet in depth. Keep up the good work my friend!

postgresql has generate_series function which can be used to create counters directly :)

why are live objects unmarked after one phase as they can continue to be live in the next phase as well ?

Thank you for doing this!

Arpit Most modern systems today follow a non-contradictory design approach. Examples are like Uber, Swiggy, and Flipkart. I cannot think of real time example of spiral designs I saw in office in any of my orgs where I worked on. I will think more on this :)

It has a huge application in blockchain, cosmos sdk heavily uses embedded databases

Amazing efforts to provide high level of dynamoDB paper. Engaging video overall. Is there any plans for MongoDB?

Database resilience 25:00

I Like the concept of hockey spike when dhoni is on field Intresting 21:46

Nobody talks about these important bits on which the systems are built up on. Good to hear someone talk about them.

I don't understand this part: @19:25. You say: "Your same event could be consumed by multiple services" How? (in terms of Kafka) From what I understood, for a topic you'll have a consumer group, but an event will go a particular partition where only 1 consumer can listen to that. (That's what Kafka does, right?) So how does this statement stands "Your same event could be consumed by multiple services"? What am I missing? Please help!!

Hey there Arpit! Amazing presentation. Is there a way we can download the notes in some kind of pdf if available?

when we provision/shift to green infra, is DB also replicated generally? Or its same for both instances - considering there can be schema/migration changes edit - had commented mid way, turns out DB is covered in limitations :) thanks!

in chrome/windows, I am having to manually (or using shortcut) activate hypothes-is on each tab, any solution? thanks

What note taking app and pen are you using for this setup ?

why to write in mySQL? why couldn't scheduler read from kinesis to retry?