From 335230f4f343724d6a5816c2e8489450960289bd Mon Sep 17 00:00:00 2001
From: Chris Lu <chrislusf@users.noreply.github.com>
Date: Sun, 28 Feb 2021 18:51:53 -0800
Subject: [PATCH] Created Async Backup (markdown)

---
 Async-Backup.md | 82 +++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 82 insertions(+)
 create mode 100644 Async-Backup.md

diff --git a/Async-Backup.md b/Async-Backup.md
new file mode 100644
index 0000000..4394eb9
--- /dev/null
+++ b/Async-Backup.md
@@ -0,0 +1,82 @@
+Cloud storage options, such as Amazon S3, Google Cloud Storage, Azure, Backblaze B2, etc, are ideal for backup purpose.
+
+For example, for Amazon S3, the upload is free. You only pay for the storage.
+So you have the benefit of:
+* Extremely fast access to local SeaweedFS Filer
+* Near-Real-Time Backup to Amazon S3 with zero-cost upload network traffic.
+
+Of course, you can also backup to local disks on another machine.
+
+# Architecture
+
+```
+  Filer --> Metadata Change Logs --> `weed filer.backup` --> AWS S3
+                                                     |
+                                                     +-----> GCP
+                                                     |
+                                                     +-----> Azure
+                                                     |
+                                                     +-----> Backblaze B2
+                                                     |
+                                                     +-----> Local Disk
+```
+
+All file meta data changes in Filer are saved in the logs and can be subscribed. See [[Filer Change Data Capture]].
+A "weed filer.backup" process will subscribe to this topic, and then read the actual file content, and send the update to the cloud sink or local disk sinks.
+
+* Sinks can be: AWS S3, Google Cloud Storage, Microsoft Azure, Backblaze B2, or Local Disk.
+
+
+# Configuration
+
+This command replaced the previous `weed filer.replicate`, which requires an external message queue.
+But for configuration, use the same `weed scaffold -config=replication` to generate a `replication.toml` file. Just need to keep the linkes of the sinks that you want to use.
+
+```
+[sink.s3]
+# read credentials doc at https://docs.aws.amazon.com/sdk-for-go/v1/developer-guide/sessions.html
+# default loads credentials from the shared credentials file (~/.aws/credentials).
+enabled = false
+aws_access_key_id     = ""     # if empty, loads from the shared credentials file (~/.aws/credentials).
+aws_secret_access_key = ""     # if empty, loads from the shared credentials file (~/.aws/credentials).
+region = "us-east-2"
+bucket = "backupbucket"        # an existing bucket
+directory = "/"                # destination directory
+endpoint = "http://localhost:8334"
+is_incremental = false
+
+```
+
+# Running Backup
+1. Make sure the `replication.toml` is in place.
+1. Start the backup by running `weed filer.backup`.
+
+# Incremental Mode
+If `is_incremental = true`, all the files are backed up under the `YYYY-MM-DD` directories, which the timestamps are based on modified time.
+So 
+* Each date directory contains all new and updated files. 
+* The deleted files in the source filer will not be deleted on the backup.
+
+So if in this folder, on `2021-03-01`, these files are created in the source:
+```
+  /dir1/file1
+  /dir1/file2
+  /dir1/file3
+```
+and on `2021-03-02`, these files are created, modified, deleted in the source:
+```
+  /dir1/file1 // modified
+  /dir1/file2 // not changed
+  /dir1/file3 // deleted
+  /dir1/file4 // created
+```
+
+The backup destination will have the following directory structure.
+```
+  /2021-03-01/dir1/file1
+  /2021-03-01/dir1/file2
+  /2021-03-01/dir1/file3
+  /2021-03-02/dir1/file1
+  /2021-03-02/dir1/file4
+```
+