iskm/learnxinyminutes-docs
1
0
Fork 0
mirror of https://github.com/adambard/learnxinyminutes-docs.git synced 2024-12-23 09:41:36 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity

[zfs/en] Update (#4817)

Browse Source
This commit is contained in:
Prayag Bhakar 2024-05-13 02:46:20 -04:00 committed by GitHub
parent 5ffa5d8356
commit bd36d7714b
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 92 additions and 22 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

114
zfs.html.markdown
View File

@ -3,39 +3,38 @@ category: tool
tool: zfs
contributors:
- ["sarlalian", "http://github.com/sarlalian"]
- ["81reap", "https://github.com/81reap"]
- ["A1EF", "https://github.com/A1EF"]
filename: LearnZfs.txt
---
[ZFS](http://open-zfs.org/wiki/Main_Page)
is a rethinking of the storage stack, combining traditional file systems as well as volume
managers into one cohesive tool. ZFS has some specific terminology that sets it apart from
more traditional storage systems, however it has a great set of features with a focus on
usability for systems administrators.
## ZFS Concepts
### Virtual Devices
A VDEV is similar to a raid device presented by a RAID card, there are several different
types of VDEV's that offer various advantages, including redundancy and speed. In general
VDEV's offer better reliability and safety than a RAID card. It is discouraged to use a
RAID setup with ZFS, as ZFS expects to directly manage the underlying disks.
A VDEV (Virtual Device) in ZFS is analogous to a RAID device and simmilaly offers different
benefits in terms of redundancy and performance. In general VDEV's offer better reliability
and safety than a RAID card. It is discouraged to use a RAID setup with ZFS, as ZFS expects
to directly manage the underlying disks.
Types of VDEV's
| VDEV Type | Similar RAID | Notes |
|-----------|----------------|---------------------------------------|
| Mirror | RAID 1 | Supports n-way mirroring for redundancy. |
| raidz1 | RAID 5 | Single disk parity, offering fault tolerance of one disk failure. |
| raidz2 | RAID 6 | Two-disk parity, can tolerate two disk failures. |
| raidz3 | - | Three-disk parity, can tolerate three disk failures. |
| Disk | - | Represents a single physical disk in a VDEV. |
| File | - | File-based VDEV, not recommended for production as it adds complexity and reduces reliability. |
* mirror (n-way mirrors supported)
* raidz
* raidz1 (1-disk parity, similar to RAID 5)
* raidz2 (2-disk parity, similar to RAID 6)
* raidz3 (3-disk parity, no RAID analog)
* disk
* file (not recommended for production due to another filesystem adding unnecessary layering)
Your data is striped across all the VDEV's present in your Storage Pool, so more VDEV's will
increase your IOPS.
Data in a ZFS storage pool is striped across all VDEVs. Adding more VDEVs, Logs, or Caches
can increase IOPS (Input/Output Operations Per Second), enhancing performance. It's crucial
to balance VDEVs for optimal performance and redundancy.
### Storage Pools
@ -48,14 +47,12 @@ ZFS datasets are analogous to traditional filesystems but with many more feature
provide many of ZFS's advantages. Datasets support [Copy on Write](https://en.wikipedia.org/wiki/Copy-on-write)
snapshots, quota's, compression and de-duplication.
### Limits
One directory may contain up to 2^48 files, up to 16 exabytes each. A single storage pool
can contain up to 256 zettabytes (2^78) of space, and can be striped across 2^64 devices. A
single host can have 2^64 storage pools. The limits are huge.
## Commands
### Storage Pools
@ -144,7 +141,6 @@ Remove zpool
$ zpool destroy test
```
### Datasets
Actions:
@ -258,6 +254,82 @@ zroot/var none none
...
```
### Write Log Pool
The ZFS Intent Log (ZIL) is a write log designed to speed up syncronus writes. This is
typically a faster drive or drive partition than the larger storage pools.
```bash
# Add a log pool
$ zpool add mypool/lamb log /dev/sdX
# Check the configureation
$ zpool status mypool/lamb
```
### Read Cache Pool
The Level 2 Adaptive Replacement Cache (L2ARC) extends the primary ARC (in-RAM cache) and is
used for read caching. This is typically a faster drive or drive partition than the larger
storage pools.
```bash
# Add a cache pool
$ zpool add mypool/lamb cache /dev/sdY
# Check the configureation
$ zpool status mypool/lamb
```
### Data Compression
Data compression reduces the amount of space data occupies on disk in excange for some extra
CPU usage. When enabled, it can enhance performance by reducing the amount of disk I/O. It
especially beneficial on systems with more CPU resources than disk bandwidth.
```bash
# Get compression options
$ zfs get -help
...
compression NO YES on | off | lzjb | gzip | gzip-[1-9] | zle | lz4 | zstd | zstd-[1-19] | zstd-fast | zstd-fast-[1-10,20,30,40,50,60,70,80,90,100,500,1000]
...
# Set compression
$ zfs set compression=on mypool/lamb
# Check the configureation
$ zpool get compression mypool/lamb
```
### Encryption at Rest
Encryption allows data to be encrypted on the device at the cost of extra CPU cycles. This
propery can only be set when a dataset is being created.
```bash
# Enable encryption on the pool
$ zpool set feature@encryption=enabled black_hole
# Create an encrypted dataset with a prompt
$ zfs create -o encryption=on -o keyformat=passphrase black_hole/enc
# Check the configureation
$ zfs get encryption black_hole/enc
```
It should be noted that there are parts of the system where the data is not encrypted. See
the table below for a breakdown.
| Component | Encrypted | Notes |
|----------------------|-------------------------------------------|------------------------------------------------------|
| Main Data Storage | Yes | Data in datasets/volumes is encrypted. |
| ZFS Intent Log (ZIL) | Yes | Synchronous write requests are encrypted. |
| L2ARC (Cache) | Yes | Cached data is stored in an encrypted form. |
| RAM (ARC) | No | Data in the primary ARC, in RAM, is not encrypted. |
| Swap Area | Conditional | Encrypted if the ZFS swap dataset is encrypted. |
| ZFS Metadata | Yes | Metadata is encrypted for encrypted datasets. |
| Snapshot Data | Yes | Snapshots of encrypted datasets are also encrypted. |
| ZFS Send/Receive | Conditional | Encrypted during send/receive if datasets are encrypted and `-w` flag is used. |
### Snapshots
@ -277,7 +349,6 @@ Actions:
* Send / Receive
* Clone
Create snapshots
```bash
@ -392,7 +463,6 @@ echo "STOP SLAVE;" | /usr/local/bin/mysql -u root -pmyrootpassword -h staging
echo "RESET SLAVE;" | /usr/local/bin/mysql -u root -pmyrootpassword -h staging
```
### Additional Reading
* [BSDNow's Crash Course on ZFS](http://www.bsdnow.tv/tutorials/zfs)