#bitmapindexes #databseindexes #bitmaps #systemdesign #learnsystemdesign #indexing In this video learn how bitmaps can be used to create database indexing in conjunction to B+ trees or AVL trees
Great tutorial! Thank for covering variety of topics. It would ge great if you start putting notes, slides, references to a Github repository. In this way, other developers can also contribute.
waow .. first time I opened this video.. It really made me fall in love with your teaching skills, way of communication and your kind body language ❤❤🔥🔥🔥🔥🔥.. Your are legend sir 👍👍👍
Question: What if any record from table is deleted ? Just turning that bit to 0 would not work much, as query could be number of users who are not approved, which in turn can include deleted users as well
If we create an index like CREATE INDEX index_name ON table_name (column_name). will the database handle it automatically for BITMAP or do I have to do any extra things to use BITMAP indexing for status column?
May be the table needs a better design here. Why not pull out status and stock in their own tables and use the status_id and stock_id as columns in the main table. Integer comparisons are faster.
What you described as "Cardinality" is what I learned as "data sparsity". Cardinality refers to the relationship between rows in different tables, e.g. 1 to many (1:M), 1 to 1 (1:1) or many to many (M:N).
Hi, thanks for the video, how will the update work, if I have to update the nth element from 1 to 0, how would I seek that element without having to go through all the numbers. Also having ids in order seem to be one of the biggest requirement for this use-case, which is not possible most times.
Conventional wisdom hulds that bitmap indexes are most appropriate for columns having low distinct values--such as GENDER, MARITAL_STATUS, and RELATION. This assumption is not completely accurate, however. In reality, a bitmap index is always advisable for systems in which data is not frequently updated by many concurrent systems. In fact, as I'll demonstrate here, a bitmap index on a culumn with 100-percent unique values (a culumn candidate for primary key) is as efficient as a B-tree index. Usage of bitmap indexes is not in fact cardinality dependent but rather application dependent. SRC : www.oracle.com/technical-resources/articles/sharma-indexes.html
Thanks for sharing. However this doesn't answer the following. 1) How is the bit map index updated when you add a new ID as well as add a new type called "ON HOLD". Is the index resized when it crosses a particular threshold (say 2^n -1 )? Is it a power of 2? 2) How is the complexity of the query reduced? For e.g select all products where status in ALL except rejected and stock in ALL. In this case, how are we saving on compute because we need to loop through all bitmaps on both columns. If that's the case, there's no exploitation of the fact that the cardinality of this column in reduced. For example. why not compute a bitmap for TRUE, FALSE, TRUE || FALSE, TRUE && FALSE and keep it ready as compared to just TRUE , FALSE now. We can afford to do this as the cardinality is less.
In theory, yes, there are only 3 distinct values in this column, and yes, that means that one only really needs two bits to hold that info. But if we continue down that route, we now have a column that requires just two bits. How can one generalise that into anything useful? How can it be efficiently stored and indexed? What does one do when someone adds some new statuses? It’s just not practical. Using bitmap indexing where appropriate is a useful and good generalisation.
Omg you could explain everything in 5 minutes instead of repeating 100 times each simple idea. And you didn't explain why it's better than seq scan, great!! (no)
Great tutorial! Thank for covering variety of topics. It would ge great if you start putting notes, slides, references to a Github repository. In this way, other developers can also contribute.
Thank you for very good explaination.
thank you for explaining the topic in such a simple and effective manner.
U explains things very clearly,love your content
Thank you, sir! You saved my day! Nice explanation 🙏
Great video! Simple example to explain complex concept - love it! Thanks. : )
Excellent explaination in a very simpler way,thank you
Thanks for sharing. Good material
good explanation. better than documentations I read before
waow .. first time I opened this video.. It really made me fall in love with your teaching skills, way of communication and your kind body language ❤❤🔥🔥🔥🔥🔥.. Your are legend sir 👍👍👍
Congrats bro , I opened it for the 100th time today
Will continue opening it everyday till I reach my grave
Great explanation, Redis provides bitmap data structure.
Hi, nice vídeo!
I would like to have it short, but without losing any content.
Thanks for sharing 👍
Excellent explanation! Thank You!
Thanks from Malaysia. It is really clear.
Great explanation, thanks for sharing knowledge, My only suggestion is to reduce repetitive statements.
Great Job Broo! The way you speak and explain is very clear!
Thank you for a very good tutorial.
thank you so much sir your explanation is awesome
Great explanation! Thanks so much!
Well explained. Thank you!
Good One!
amazing ;)
Super video. Thanks
Your videos are amazing
Thanks for this video!!! Man!!! Could u do a video about an example using bitmap and b-tree together :D Have a nice day!
Hey, it will be great if you also share the reference materials as well. thanks
Hey, you explained Bitmap indexing concept really well 👍🏻. Do you have one for cluster indexing?
question: if you have to go row by row in the bitmap index to find the matches, how is it different than a fullscan?
Best explaination ever
Thanks You!
As for as oracle , the definition for cardinality is some what not same as you explained , but your explanation of bitmap is good
thank you sir. it is really helpful. could you please suggest me a reference paper or book for this?
How about to have a session discussing about the design distributed message queue system?
thanks
Thank u very much
Good One
hey! can you do a system design for Expidia?
crystal clear
Question: What if any record from table is deleted ? Just turning that bit to 0 would not work much, as query could be number of users who are not approved, which in turn can include deleted users as well
Even if we find all 1st in bitwise array, it's O(N) ?
What do you think of CrimsonDB?
Do some live with us!!!!!
If we create an index like CREATE INDEX index_name ON table_name (column_name). will the database handle it automatically for BITMAP or do I have to do any extra things to use BITMAP indexing for status column?
May be the table needs a better design here. Why not pull out status and stock in their own tables and use the status_id and stock_id as columns in the main table. Integer comparisons are faster.
can you suggest the list of books plz?
Isn’t this same as columnar format the parquet uses?
Good content, but the volume is very low.
What is the relationship between bitmap indexing and high dimensional indexing problem???
Hi, bro, I see you hadn't updated for nearly half a year. Are you OK?
What you described as "Cardinality" is what I learned as "data sparsity". Cardinality refers to the relationship between rows in different tables, e.g. 1 to many (1:M), 1 to 1 (1:1) or many to many (M:N).
'cardinality' is the same as 'size'
en.wikipedia.org/wiki/Cardinal_number
sparsity describes the number of sparse items (e.g. empty rows in DB)
Hi, thanks for the video, how will the update work, if I have to update the nth element from 1 to 0, how would I seek that element without having to go through all the numbers. Also having ids in order seem to be one of the biggest requirement for this use-case, which is not possible most times.
Great Explanation! so you explained it nicely for 6 rows and it has 6 bits but how does it scale for millions of rows. we can't have millions of bits.
Whats' the issues with millions rows. If we you take 1 million rows and 4 cardinality, its just 500KB data.
Conventional wisdom hulds that bitmap indexes are most appropriate for columns having low distinct values--such as GENDER, MARITAL_STATUS, and RELATION. This assumption is not completely accurate, however. In reality, a bitmap index is always advisable for systems in which data is not frequently updated by many concurrent systems. In fact, as I'll demonstrate here, a bitmap index on a culumn with 100-percent unique values (a culumn candidate for primary key) is as efficient as a B-tree index.
Usage of bitmap indexes is not in fact cardinality dependent but rather application dependent.
SRC : www.oracle.com/technical-resources/articles/sharma-indexes.html
Thanks for sharing. However this doesn't answer the following.
1) How is the bit map index updated when you add a new ID as well as add a new type called "ON HOLD". Is the index resized when it crosses a particular threshold (say 2^n -1 )? Is it a power of 2?
2) How is the complexity of the query reduced? For e.g select all products where status in ALL except rejected and stock in ALL. In this case, how are we saving on compute because we need to loop through all bitmaps on both columns.
If that's the case, there's no exploitation of the fact that the cardinality of this column in reduced. For example. why not compute a bitmap for TRUE, FALSE, TRUE || FALSE, TRUE && FALSE and keep it ready as compared to just TRUE , FALSE now. We can afford to do this as the cardinality is less.
actually this video is useless =/
I too had the same questions in mind.
Why can't we use 3 numbers for each row. 0 for approved, 1 for pending, 3 for rejected.
It is a bit, not character/string datatype
In theory, yes, there are only 3 distinct values in this column, and yes, that means that one only really needs two bits to hold that info. But if we continue down that route, we now have a column that requires just two bits. How can one generalise that into anything useful? How can it be efficiently stored and indexed? What does one do when someone adds some new statuses? It’s just not practical. Using bitmap indexing where appropriate is a useful and good generalisation.
Anyone knows his twitter handle, please let me know
Omg you could explain everything in 5 minutes instead of repeating 100 times each simple idea. And you didn't explain why it's better than seq scan, great!! (no)
Indicies, not indexes. Thought you were all geniuses?
thanks