maokaihe/mini_lsm
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

docs: s/tutorial/course

Browse Source 
Signed-off-by: Alex Chi Z <iskyzh@gmail.com>
This commit is contained in:
Alex Chi Z
2025-01-19 19:28:04 -05:00
parent 7f4b204064
commit 38c7ef6f2f
32 changed files with 67 additions and 67 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
mini-lsm-book/src/week1-06-write-path.md
View File

@@ -21,7 +21,7 @@ cargo x scheck
## Task 1: Flush Memtable to SST
At this point, we have all in-memory things and on-disk files ready, and the storage engine is able to read and merge the data from all these structures. Now, we are going to implement the logic to move things from memory to the disk (so-called flush), and complete the Mini-LSM week 1 tutorial.
At this point, we have all in-memory things and on-disk files ready, and the storage engine is able to read and merge the data from all these structures. Now, we are going to implement the logic to move things from memory to the disk (so-called flush), and complete the Mini-LSM week 1 course.
In this task, you will need to modify:
@@ -36,7 +36,7 @@ You will need to modify `LSMStorageInner::force_flush_next_imm_memtable` and `Me
* Create an SST file corresponding to a memtable.
* Remove the memtable from the immutable memtable list and add the SST file to L0 SSTs.
We have not explained what is L0 (level-0) SSTs for now. In general, they are the set of SSTs files directly created as a result of memtable flush. In week 1 of this tutorial, we will only have L0 SSTs on the disk. We will dive into how to organize them efficiently using leveled or tiered structure on the disk in week 2.
We have not explained what is L0 (level-0) SSTs for now. In general, they are the set of SSTs files directly created as a result of memtable flush. In week 1 of this course, we will only have L0 SSTs on the disk. We will dive into how to organize them efficiently using leveled or tiered structure on the disk in week 2.
Note that creating an SST file is a compute-heavy and a costly operation. Again, we do not want to hold the `state` read/write lock for a long time, as it might block other operations and create huge latency spikes in the LSM operations. Also, we use the `state_lock` mutex to serialize state modification operations in the LSM tree. In this task, you will need to think carefully how to use these locks to make the LSM state modification race-condition free while minimizing critical sections.