meta data for this page
Differences
This shows you the differences between two versions of the page.
| Both sides previous revisionPrevious revisionNext revision | Previous revisionLast revisionBoth sides next revision | ||
| zfs [2015/03/23 14:31] – niziak | linux:fs:zfs [2021/03/02 12:59] – niziak | ||
|---|---|---|---|
| Line 1: | Line 1: | ||
| ====== ZFS ====== | ====== ZFS ====== | ||
| - | ===== Creating ZFS dataset | + | [[https:// |
| + | |||
| + | Features: | ||
| + | * data pools (tanks) are abstraction aggregate block devices (simple, mirror, raidz, spares, etc) | ||
| + | * data set is created on data pool or another (parent) data set. | ||
| + | * whole data pool space is shared between | ||
| + | * compression | ||
| + | * block level deduplication (not usable for emails with attachment, where attachment are shifted to different offset) | ||
| + | |||
| + | OpenZFS2.0.0 (Dec 20) [[https:// | ||
| + | * Sequential resilver (rebuild only used by data portions) | ||
| + | * Persistent L2ARC cache (survives between reboots) | ||
| + | * ZSTD | ||
| + | * Redacted replication (replicate with some data excluded) | ||
| + | * FreeBSD and Linux unification | ||
| + | |||
| + | Proposed use case: | ||
| + | POOL created on encrypted LUKS block device. | ||
| < | < | ||
| + | POOL | ||
| + | |-- /filer (quota) | ||
| + | | ||
| + | | ||
| + | | ||
| + | | ||
| + | | | ||
| + | |-- /home (compression, | ||
| + | +-- /var (quota) | ||
| + | +- log (compression) | ||
| + | </ | ||
| + | |||
| + | ===== ZFS implementations ===== | ||
| + | |||
| + | ZFS-Fuse 0.7 is using old pool version 23, where [[http:// | ||
| + | [[http:// | ||
| + | |||
| + | |||
| + | |||
| + | ===== Creating ZFS dataset ===== | ||
| + | |||
| + | <code bash> | ||
| zpool create INBOX /dev/loop0 /dev/loop1 /dev/loop2 /dev/loop3 | zpool create INBOX /dev/loop0 /dev/loop1 /dev/loop2 /dev/loop3 | ||
| </ | </ | ||
| - | < | + | < |
| # zpool list | # zpool list | ||
| NAME SIZE ALLOC | NAME SIZE ALLOC | ||
| Line 13: | Line 52: | ||
| </ | </ | ||
| - | < | + | < |
| # zpool status | # zpool status | ||
| pool: INBOX | pool: INBOX | ||
| Line 32: | Line 71: | ||
| Dataset " | Dataset " | ||
| - | < | + | < |
| # zfs list | # zfs list | ||
| NAME USED AVAIL REFER MOUNTPOINT | NAME USED AVAIL REFER MOUNTPOINT | ||
| Line 55: | Line 94: | ||
| zpool add INBOX /dev/loop4 | zpool add INBOX /dev/loop4 | ||
| </ | </ | ||
| - | ===== Deduplication and compression ===== | ||
| - | Enable compression and dedupliaction in parent dataset (will be inherited by childs) | + | ===== Deduplication ===== |
| < | < | ||
| zfs set dedup=on INBOX | zfs set dedup=on INBOX | ||
| - | zfs set compression=on INBOX | ||
| </ | </ | ||
| + | New attributed applies only to newly written data. | ||
| - | But new attributed applies only to newly written | + | Tests |
| + | For test I was using 3 files 16MB each of random | ||
| + | Above 3 files takes 38,6M on disc: | ||
| + | < | ||
| + | # zdb -S INBOX | ||
| + | Simulated DDT histogram: | ||
| + | bucket | ||
| + | ______ | ||
| + | refcnt | ||
| + | ------ | ||
| + | | ||
| + | | ||
| + | |||
| + | dedup = 1.00, compress = 1.00, copies = 1.00, dedup * compress / copies = 1.00 | ||
| + | </ | ||
| + | |||
| + | Additionaly one big file with content B1|B2|B3 was added to filesystem: | ||
| < | < | ||
| - | NAME SIZE ALLOC | + | # zdb -S INBOX |
| - | INBOX | + | Simulated DDT histogram: |
| + | |||
| + | bucket | ||
| + | ______ | ||
| + | refcnt | ||
| + | ------ | ||
| + | | ||
| + | | ||
| + | |||
| + | dedup = 2.00, compress = 1.00, copies = 1.00, dedup * compress / copies = 2.00 | ||
| </ | </ | ||
| - | The same data copied again to dataset after compression enabled | + | Additionaly one big file with content B1|B2|B3|B1|B2|B3 was added to filesystem: |
| + | < | ||
| + | # zdb -S INBOX | ||
| + | Simulated DDT histogram: | ||
| + | |||
| + | bucket | ||
| + | ______ | ||
| + | refcnt | ||
| + | ------ | ||
| + | | ||
| + | | ||
| + | |||
| + | dedup = 4.00, compress = 1.00, copies = 1.00, dedup * compress / copies = 4.00 | ||
| + | |||
| + | </ | ||
| + | Next, new file with content 0|B1|B2|B3 (one dummy byte plus B1|B2|B3) was added: | ||
| < | < | ||
| - | NAME SIZE ALLOC | + | # zdb -S INBOX |
| - | INBOX | + | Simulated DDT histogram: |
| + | |||
| + | bucket | ||
| + | ______ | ||
| + | refcnt | ||
| + | ------ | ||
| + | | ||
| + | | ||
| + | Total 769 | ||
| + | |||
| + | dedup = 2.50, compress = 1.00, copies = 1.00, dedup * compress / copies = 2.50 | ||
| </ | </ | ||
| - | ====== Summary: ====== | + | **So ZFS cannot match shifted data and make deduplication!** |
| - | Filesystem | + | Additional simple test. Two files: |
| + | |0|B1|0|B2|0|B3|0| | ||
| + | |0|B1|B2|B3| | ||
| + | Only beginning of both files |0|B1| | ||
| - | SquashFS=365MB vs ZFS=563MB | + | ZFS provides block level deduplication based on block checksums which we got almost for free. |
| - | Deduplication on file levels works on ZFS and SquashFS (the same folder copied again). | ||
| - | ====== References: ====== | + | ===== Compression ===== |
| + | |||
| + | Enable compression and dedupliaction in parent dataset (will be inherited by childs) | ||
| + | |||
| + | < | ||
| + | zfs set compression=on INBOX | ||
| + | </ | ||
| + | Possible parameters for compression=on | off | lzjb | gzip | gzip-[1-9] | zle | ||
| + | New attributed applies only to newly written data. For test data I was using Maildir with some huge e-mails. | ||
| + | |||
| + | ^compression ^ logical size ^ physical size^ ratio | | ||
| + | |off | ||
| + | |on = lzjb | ||
| + | |gzip-1 | ||
| + | |gzip=gzip-6 | 702 MB | 359 MB | 1.95 | | ||
| + | |gzip-9 | ||
| + | |-- | ||
| + | |squashfs | ||
| + | |||
| + | |||
| + | zdb -S INBOX | ||
| + | zdb -b INBOX | ||
| + | |||
| + | |||
| + | |||
| + | <code bash> | ||
| + | zfs get compressratio | ||
| + | </ | ||
| + | |||
| + | ===== References: ===== | ||
| + | [[http:// | ||
| + | [[https:// | ||
| [[http:// | [[http:// | ||
| + | |||
| + | [[http:// | ||
| + | |||
| + | [[http:// | ||