Gentoo Archives: gentoo-user

From: covici@××××××××××.com
To: gentoo-user@l.g.o
Subject: Re: [gentoo-user] ZFS
Date: Tue, 17 Sep 2013 16:32:39
Message-Id: 20167.1379435549@ccs.covici.com
In Reply to: Re: [gentoo-user] ZFS by Pandu Poluan
1 Pandu Poluan <pandu@××××××.info> wrote:
2
3 > On Tue, Sep 17, 2013 at 2:20 PM, Grant <emailgrant@×××××.com> wrote:
4 > > I'm convinced I need 3-disk RAID1 so I can lose 2 drives and keep
5 > > running. I'd also like to stripe for performance, resulting in
6 > > RAID10. It sounds like most hardware controllers do not support
7 > > 6-disk RAID10 so ZFS looks very interesting.
8 > >
9 > > Can I operate ZFS RAID without a hardware RAID controller?
10 > >
11 >
12 > Yes. In fact, that's ZFS' preferred mode of operation (i.e., it
13 > handles all redundancy by itself).
14 >
15 > > From a RAID perspective only, is ZFS a better choice than conventional
16 > > software RAID?
17 > >
18 >
19 > Yes.
20 >
21 > ZFS checksummed all blocks during writes, and verifies those checksums
22 > during read.
23 >
24 > It is possible to have 2 bits flipped at the same time among 2 hard
25 > disks. In such case, the RAID controller will never see the bitflips.
26 > But ZFS will see it.
27 >
28 > > ZFS seems to have many excellent features and I'd like to ease into
29 > > them slowly (like an old man into a nice warm bath). Does ZFS allow
30 > > you to set up additional features later (e.g. snapshots, encryption,
31 > > deduplication, compression) or is some forethought required when first
32 > > making the filesystem?
33 > >
34 >
35 > Snapshots is built-in from the beginning. All you have to do is create
36 > one when you want it.
37 >
38 > Deduplication can be turned on and off at will -- but be warned: You
39 > need HUGE amount of RAM.
40 >
41 > Compression can be turned on and off at will. Previously-compressed
42 > data won't become uncompressed unless you modify them.
43 >
44 > > It looks like there are comprehensive ZFS Gentoo docs
45 > > (http://wiki.gentoo.org/wiki/ZFS) but can anyone tell me from the real
46 > > world about how much extra difficulty/complexity is added to
47 > > installation and ongoing administration when choosing ZFS over ext4?
48 > >
49 >
50 > Very very minimal. So minimal, in fact, that if you don't plan to use
51 > ZFS as a root filesystem, it's laughably simple. You don't even have
52 > to edit /etc/fstab
53 >
54 > > Performance doesn't seem to be one of ZFS's strong points. Is it
55 > > considered suitable for a high-performance server?
56 > >
57 > > http://www.phoronix.com/scan.php?page=news_item&px=MTM1NTA
58 > >
59 >
60 > Several points:
61 >
62 > 1. The added steps of checksumming (and verifying the checksums)
63 > *will* give a performance penalty.
64 >
65 > 2. When comparing performance of 1 (one) drive, of course ZFS will
66 > lose. But when you build a ZFS pool out of 3 pairs of mirrored drives,
67 > throughput will increase significantly as ZFS has the ability to do
68 > 'load-balancing' among mirror-pairs (or, in ZFS parlance, "mirrored
69 > vdevs")
70 >
71 > Go directly to this post:
72 > http://phoronix.com/forums/showthread.php?79922-Benchmarks-Of-The-New-ZFS-On-Linux-EXT4-Wins&p=326838#post326838
73 >
74 > Notice how ZFS won against ext4 in 8 scenarios out of 9. (The only
75 > scenario where ZFS lost is in the single-client RAID-1 scenario)
76 >
77 > > Besides performance, are there any drawbacks to ZFS compared to ext4?
78 > >
79 >
80 > 1. You need a huge amount of RAM to let ZFS do its magic. But RAM is
81 > cheap nowadays. Data... possibly priceless.
82 >
83 > 2. Be careful when using ZFS on a server on which processes rapidly
84 > spawn and terminate. ZFS doesn't like memory fragmentation.
85 >
86 > For point #2, I can give you a real-life example:
87 >
88 > My mail server, for some reasons, choke if too many TLS errors happen.
89 > So, I placed "Perdition" in to capture all POP3 connections and
90 > 'un-TLS' them. Perdition spawns a new process for *every* connection.
91 > My mail server has 2000 users, I regularly see more than 100 Perdition
92 > child processes. Many very ephemeral (i.e., existing for less than 5
93 > seconds). The RAM is undoubtedly *extremely* fragmented. ZFS cries
94 > murder when it cannot allocate a contiguous SLAB of memory to increase
95 > its ARC Cache.
96 >
97 > OTOH, on another very busy server (mail archiving server using
98 > MailArchiva, handling 2000+ emails per hour), ZFS run flawlessly. No
99 > incident _at_all_. Undoubtedly because MailArchiva use one single huge
100 > process (Java-based) to handle all transactions, so no RAM
101 > fragmentation here.
102 Spo do I need that overlay at all, or just emerge zfs and its module?
103 Also, I now have lvm volumes, including root, but not boot, how to
104 convert and do I have to do anything to my initramfs?
105
106 --
107 Your life is like a penny. You're going to lose it. The question is:
108 How do
109 you spend it?
110
111 John Covici
112 covici@××××××××××.com

Replies

Subject Author
Re: [gentoo-user] ZFS Douglas J Hunley <doug.hunley@×××××.com>
Re: [gentoo-user] ZFS Douglas J Hunley <doug.hunley@×××××.com>