1 |
On Tue, Sep 17, 2013 at 2:20 PM, Grant <emailgrant@×××××.com> wrote: |
2 |
> I'm convinced I need 3-disk RAID1 so I can lose 2 drives and keep |
3 |
> running. I'd also like to stripe for performance, resulting in |
4 |
> RAID10. It sounds like most hardware controllers do not support |
5 |
> 6-disk RAID10 so ZFS looks very interesting. |
6 |
> |
7 |
> Can I operate ZFS RAID without a hardware RAID controller? |
8 |
> |
9 |
|
10 |
Yes. In fact, that's ZFS' preferred mode of operation (i.e., it |
11 |
handles all redundancy by itself). |
12 |
|
13 |
> From a RAID perspective only, is ZFS a better choice than conventional |
14 |
> software RAID? |
15 |
> |
16 |
|
17 |
Yes. |
18 |
|
19 |
ZFS checksummed all blocks during writes, and verifies those checksums |
20 |
during read. |
21 |
|
22 |
It is possible to have 2 bits flipped at the same time among 2 hard |
23 |
disks. In such case, the RAID controller will never see the bitflips. |
24 |
But ZFS will see it. |
25 |
|
26 |
> ZFS seems to have many excellent features and I'd like to ease into |
27 |
> them slowly (like an old man into a nice warm bath). Does ZFS allow |
28 |
> you to set up additional features later (e.g. snapshots, encryption, |
29 |
> deduplication, compression) or is some forethought required when first |
30 |
> making the filesystem? |
31 |
> |
32 |
|
33 |
Snapshots is built-in from the beginning. All you have to do is create |
34 |
one when you want it. |
35 |
|
36 |
Deduplication can be turned on and off at will -- but be warned: You |
37 |
need HUGE amount of RAM. |
38 |
|
39 |
Compression can be turned on and off at will. Previously-compressed |
40 |
data won't become uncompressed unless you modify them. |
41 |
|
42 |
> It looks like there are comprehensive ZFS Gentoo docs |
43 |
> (http://wiki.gentoo.org/wiki/ZFS) but can anyone tell me from the real |
44 |
> world about how much extra difficulty/complexity is added to |
45 |
> installation and ongoing administration when choosing ZFS over ext4? |
46 |
> |
47 |
|
48 |
Very very minimal. So minimal, in fact, that if you don't plan to use |
49 |
ZFS as a root filesystem, it's laughably simple. You don't even have |
50 |
to edit /etc/fstab |
51 |
|
52 |
> Performance doesn't seem to be one of ZFS's strong points. Is it |
53 |
> considered suitable for a high-performance server? |
54 |
> |
55 |
> http://www.phoronix.com/scan.php?page=news_item&px=MTM1NTA |
56 |
> |
57 |
|
58 |
Several points: |
59 |
|
60 |
1. The added steps of checksumming (and verifying the checksums) |
61 |
*will* give a performance penalty. |
62 |
|
63 |
2. When comparing performance of 1 (one) drive, of course ZFS will |
64 |
lose. But when you build a ZFS pool out of 3 pairs of mirrored drives, |
65 |
throughput will increase significantly as ZFS has the ability to do |
66 |
'load-balancing' among mirror-pairs (or, in ZFS parlance, "mirrored |
67 |
vdevs") |
68 |
|
69 |
Go directly to this post: |
70 |
http://phoronix.com/forums/showthread.php?79922-Benchmarks-Of-The-New-ZFS-On-Linux-EXT4-Wins&p=326838#post326838 |
71 |
|
72 |
Notice how ZFS won against ext4 in 8 scenarios out of 9. (The only |
73 |
scenario where ZFS lost is in the single-client RAID-1 scenario) |
74 |
|
75 |
> Besides performance, are there any drawbacks to ZFS compared to ext4? |
76 |
> |
77 |
|
78 |
1. You need a huge amount of RAM to let ZFS do its magic. But RAM is |
79 |
cheap nowadays. Data... possibly priceless. |
80 |
|
81 |
2. Be careful when using ZFS on a server on which processes rapidly |
82 |
spawn and terminate. ZFS doesn't like memory fragmentation. |
83 |
|
84 |
For point #2, I can give you a real-life example: |
85 |
|
86 |
My mail server, for some reasons, choke if too many TLS errors happen. |
87 |
So, I placed "Perdition" in to capture all POP3 connections and |
88 |
'un-TLS' them. Perdition spawns a new process for *every* connection. |
89 |
My mail server has 2000 users, I regularly see more than 100 Perdition |
90 |
child processes. Many very ephemeral (i.e., existing for less than 5 |
91 |
seconds). The RAM is undoubtedly *extremely* fragmented. ZFS cries |
92 |
murder when it cannot allocate a contiguous SLAB of memory to increase |
93 |
its ARC Cache. |
94 |
|
95 |
OTOH, on another very busy server (mail archiving server using |
96 |
MailArchiva, handling 2000+ emails per hour), ZFS run flawlessly. No |
97 |
incident _at_all_. Undoubtedly because MailArchiva use one single huge |
98 |
process (Java-based) to handle all transactions, so no RAM |
99 |
fragmentation here. |
100 |
|
101 |
|
102 |
Rgds, |
103 |
-- |
104 |
FdS Pandu E Poluan |
105 |
~ IT Optimizer ~ |
106 |
|
107 |
• LOPSA Member #15248 |
108 |
• Blog : http://pepoluan.tumblr.com |
109 |
• Linked-In : http://id.linkedin.com/in/pepoluan |