This is good, thank you. Cassandra can be used for storing something that is more of real time then, along with the things that has less relational thing then. Also, let's say we have complex analytical queries to run on data, this may not be as good because of lack of good support for indexes.
Bro, this is super useful . Please keep uploading stuff. But sooner or later you have to do the soyjack faces to go viral. You will cross 100k one day if you keep at it
So, if possible apart from the witty descriptions (which I really like lol), can you please add related research papers, link to previous video/whole playlist, related books etc too ? Would help the video quality as well as channel I think.
You've gotta expand on this question a bit. What type of reads? Analytics? Or do we just need one row at a time? A lot of indices...local, global? In theory, I'd say none and to just use snowflake or something lol.
@@jordanhasnolife5163 Yes, sure. I mean, imagine types with 200 to 300 attributes each. At least 20 attributes of each type must be searchable (filter by should be quick). In Cassandra there are secondary indices and materialized views we could use. But I guess, 20 or more indices on each of the probably 1000 different types is not the ideal use-case for cassandra. Normally we do much more queries than inserting new objects.
@@jordanhasnolife5163 Thanks. A follow-up question: Won't this limit its use case as an OLTP database? I see this is used by many companies as one of their primary sources of truth. I haven't yet seen your case-studies of various systems but won't ignoring all kinds of crazy skews make it much less effective for many scenarios.
@@jayantprakash6425 It really depends upon your use case. For example, storing a bunch of timestamped messages: why couldn't we use it for that? We aren't writing multiple messages at once, they are written once and then are immutable.
I don't actually know how dynamodb works, since it's closed source. From my understanding it's different than the dynamo paper, whereas Cassandra is the same lol
Hey Jordan - was wondering, do you know if you have to manually setup how many partitions Cassandra will support? Or will Cassandra automatically provision partitions for you if they get too full?
@@jordanhasnolife5163 very true lol. The other thing is if we have a compound partition key we need to query by the entire partition key. lastly, I get confused on when to use denormalization vs secondary indexes
For anyone like me, confused by the Data Organisation in slide 4, refer to this pic - database.guide/wp-content/uploads/2016/06/wide_column_store_database_example_row-1.png
Finally, a video that covers in-depth design, kudos
This guy really runs the extra mile to bring in-depth content
The only miles I run are on my UA-cam videos
@@jordanhasnolife5163 ha ha =D
These videos on specific tech are indeed awesome!
Thanks!
Kudos on the deep dive, especially the use cases. Very helpful for the system design interview.
Great video. It was so smart to start with all those introductions... patience pays.
thank you !
amazing content! very in-depth knowledge which most other videos lack! Thanks for teaching me, Jordan!
Omg this video is real gold and on to point.
Going to binge watch all your vids after watching this one!
Thanks Nicholas! Welcome to the channel :)
every video of yours takes me 2-3x time for me to understand. Its concise, and soo deep bro. 👊
I wish she would tell me that
@@jordanhasnolife5163 lmao
really interesting, I'm a second year student and you explain very well !
Thanks!'
This is good, thank you.
Cassandra can be used for storing something that is more of real time then, along with the things that has less relational thing then.
Also, let's say we have complex analytical queries to run on data, this may not be as good because of lack of good support for indexes.
It has solid support for indexes but generally agreed for analytical stuff you probably want to use a db built on top of parquet files
@@jordanhasnolife5163 why is that generally agreed upon then that analytical stuff is better off to something else other than Cassandra.
@@raj_kundalia Not column oriented storage
@@jordanhasnolife5163 thank you, this clears off a lot of things.
Bro, this is super useful . Please keep uploading stuff. But sooner or later you have to do the soyjack faces to go viral. You will cross 100k one day if you keep at it
Just how the internet works unfortunately, I'll keep my dry socially unnaceptable humor for now and see what happens :)
So, if possible apart from the witty descriptions (which I really like lol), can you please add related research papers, link to previous video/whole playlist, related books etc too ?
Would help the video quality as well as channel I think.
Yep, I can start doing stuff like this
great detail!
love the intro
Can you do a video on Spanner please? I would simp.
Muahahah
What NoSql database is better if you have a read oriented application with a large amount of columns and you need a lot of indices?
You've gotta expand on this question a bit. What type of reads? Analytics? Or do we just need one row at a time? A lot of indices...local, global?
In theory, I'd say none and to just use snowflake or something lol.
@@jordanhasnolife5163 Yes, sure. I mean, imagine types with 200 to 300 attributes each. At least 20 attributes of each type must be searchable (filter by should be quick).
In Cassandra there are secondary indices and materialized views we could use. But I guess, 20 or more indices on each of the probably 1000 different types is not the ideal use-case for cassandra.
Normally we do much more queries than inserting new objects.
Can you please add a reference to multiple clustering keys and multiple copies of data ?
stackoverflow.com/questions/36741042/cassandra-order-by-multiple-columns
Super useful
Hey Jordan, I couldn't find much info on isolation levels/atomicity of cassandra online. Could you elaborate on that bit?
I believe that there's just row level atomicity, and you can pretty much expect no isolation guarantees from there
@@jordanhasnolife5163 Thanks. A follow-up question: Won't this limit its use case as an OLTP database? I see this is used by many companies as one of their primary sources of truth. I haven't yet seen your case-studies of various systems but won't ignoring all kinds of crazy skews make it much less effective for many scenarios.
@@jayantprakash6425 It really depends upon your use case. For example, storing a bunch of timestamped messages: why couldn't we use it for that? We aren't writing multiple messages at once, they are written once and then are immutable.
Thank you
I can't believe this is a free course 😳😳
It's free until I need u to send me nudes
Ooofff
cassandra vs dynamo db ??
I don't actually know how dynamodb works, since it's closed source. From my understanding it's different than the dynamo paper, whereas Cassandra is the same lol
Hey Jordan - was wondering, do you know if you have to manually setup how many partitions Cassandra will support? Or will Cassandra automatically provision partitions for you if they get too full?
I think there's some automatic rebalancing due to the consistent hashing but not 100% sure
the indexes on cassandra always confuse the shit out of me.
Just think partition key, sort key :)
@@jordanhasnolife5163 very true lol. The other thing is if we have a compound partition key we need to query by the entire partition key.
lastly, I get confused on when to use denormalization vs secondary indexes
For anyone like me, confused by the Data Organisation in slide 4, refer to this pic -
database.guide/wp-content/uploads/2016/06/wide_column_store_database_example_row-1.png
Thanks for this!