From 0db82064575f2757cef64d7a641825ff494ec0eb Mon Sep 17 00:00:00 2001
From: Akira Hayakawa <ruby.wktk@gmail.com>
Date: Sat, 8 Jun 2024 09:56:36 +0900
Subject: [PATCH] doc: batched write

---
 doc/src/SUMMARY.md                            |  1 +
 .../README.md => doc/src/batched-write.md     | 33 +++++++++++++++++--
 2 files changed, 32 insertions(+), 2 deletions(-)
 rename redb-backend/README.md => doc/src/batched-write.md (64%)

diff --git a/doc/src/SUMMARY.md b/doc/src/SUMMARY.md
index c32b509c..6dc9af31 100644
--- a/doc/src/SUMMARY.md
+++ b/doc/src/SUMMARY.md
@@ -2,6 +2,7 @@
 
 - [Multi-Raft](multi-raft.md)
   - [Heartbeat Multiplexing](heartbeat-multiplexing.md)
+  - [Batched Write](batched-write.md)
 - [Raft Process](raft-process.md)
   - [Multi Threading](multi-threading.md)
   - [Application State](application-state.md)
diff --git a/redb-backend/README.md b/doc/src/batched-write.md
similarity index 64%
rename from redb-backend/README.md
rename to doc/src/batched-write.md
index e2516c74..df8eac66 100644
--- a/redb-backend/README.md
+++ b/doc/src/batched-write.md
@@ -1,4 +1,4 @@
-# redb-backend
+# Batched Write
 
 In multi-raft, multiple shards process write requests. Conceptually, each shard maintains its own log for entry insertion.
 
@@ -6,4 +6,33 @@ Having a physically independent log for each shard isn't efficient as each write
 
 However, an optimization technique called "batching" can be used. Here, each shard maintains a virtual log, and the entries are temporarily queued in a shared queue. These queued entries are then processed in a single transaction, reducing the number of transactions.
 
-This approach often presents a throughput versus latency dilemma. However, this implementation increases throughput without sacrificing latency.
\ No newline at end of file
+This approach often presents a throughput versus latency dilemma. However, this implementation increases throughput without sacrificing latency.
+
+```mermaid
+graph LR
+  CLI(Client)
+
+  subgraph P1
+    T1(redb::Table)
+  end
+  subgraph P2
+    T2(redb::Table)
+  end
+  subgraph P3
+    T3(redb::Table)
+  end
+
+  subgraph Reaper
+    Q(Queue)
+  end
+  DB[(redb::Database)]
+
+  CLI -->|entry| T1
+  CLI --> T2
+  CLI --> T3
+  T1 -->|lazy entry| Q
+  T2 --> Q
+  T3 --> Q
+  Q -->|transaction| DB
+
+```
\ No newline at end of file