1 |
Sean Cook wrote: |
2 |
|
3 |
> My post did say "Back in the Day" and it was around 1999 that we did |
4 |
> this however, then I was running unstable raid tools and they have come |
5 |
> a long way, and I was running 2.2 kernel. |
6 |
|
7 |
Hey I'm sure you saw what you saw, that's not a really an issue for me. |
8 |
However the link you posted was a study in "how not to do serious |
9 |
hardware testing." |
10 |
|
11 |
|
12 |
> However, the only people I am aware of that say the performance is |
13 |
> better on hardware raid are hardware raid manufacturers and you... You |
14 |
> don't work for LSI do you? Most of the linux software folks agree with |
15 |
> me or take a mildly more conservative tone. |
16 |
|
17 |
Oddly this is the first time I've ever heard of software RAID being |
18 |
faster than real hardware raid... I've never seen any numbers that |
19 |
remotely point to such a case in my own testing or anyone that attempted |
20 |
to do real testing. |
21 |
|
22 |
If I worked for a hardware vendor I'd hope to be getting paid more than |
23 |
I do at Internet startup #5. In any case I think you're misunderstanding |
24 |
the issues. Let's take a look at these links below. |
25 |
|
26 |
|
27 |
> Redhat Enterprise Linux (circa 2005-2006) |
28 |
> http://www.redhat.com/docs/manuals/enterprise/RHEL-4-Manual/pdf/rhel-isa-en.pdf |
29 |
> |
30 |
> Often the excess CPU power available for software RAID parity |
31 |
> calculations greatly exceeds the processing power present on a RAID |
32 |
> controller card. Therefore, some software RAID implementations |
33 |
> actually have the capability for higher performance than hardware RAID |
34 |
> implementations. |
35 |
|
36 |
This is true, but does it actually mean you get more performance? |
37 |
|
38 |
I'd argue that any self respecting RAID card has more than enough CPU |
39 |
to handle saturating the slow in comparison drives. You can argue that |
40 |
they do not and that might be valid. If you're right then that RAID card |
41 |
does indeed suck ass. |
42 |
Assuming I'm right we can say that the CPU is not a limiting factor in |
43 |
either software or hardware raid. On the other hand I can say that since |
44 |
I have dedicated processing power for my RAID that I don't have to worry |
45 |
about spikes in load causing I/O issues as well as processing issues. |
46 |
Admittedly with today's processors you'd have to be taking a gigantic |
47 |
bite out of your CPU to affect your software RAID. |
48 |
|
49 |
|
50 |
> Mysql: |
51 |
> |
52 |
> Hardware Versus Software |
53 |
> |
54 |
> Some operating systems can perform software RAID. Rather than buying a |
55 |
> dedicated RAID controller, the operating system's kernel splits the I/O |
56 |
> among multiple disks. Many users shy away from using these features |
57 |
> because they've long been considered slow or buggy. |
58 |
> |
59 |
> In reality, software RAID is quite stable and performs rather well. The |
60 |
> performance differences between hardware and software RAID tend not to |
61 |
> be significant until they're under quite a bit of load. For smaller and |
62 |
> medium-sized workloads, there's little discernible difference between |
63 |
> them. Yes, the server's CPU must do a bit more work when using software |
64 |
> RAID, but modern CPUs are so fast that the RAID operations consume a |
65 |
> small fraction of the available CPU time. And, as we stressed earlier, |
66 |
> the CPU is usually not the bottleneck in a database server anyway. |
67 |
|
68 |
I understand this to mean that software raid can be just a fast as |
69 |
hardware raid when there are no limiting factors. This is also true. |
70 |
|
71 |
Notice that neither of the above said software RAID was faster only that |
72 |
it could perform as well under the right conditions and save you money. |
73 |
I agree with that as well. |
74 |
|
75 |
However let's talk about a system that lives outside this perfect world |
76 |
that may have RAM, CPU, and I/O contention. A system where having a |
77 |
256MB cache on your RAID card turning expensive short reads and writes |
78 |
into nice long reads and writes by acting as an I/O buffer eliminates |
79 |
painful and expensive IO waits on your system. Things that you can |
80 |
actually measure in detail using iostats from the sysstats package while |
81 |
watching system load fall from 12 to .5. |
82 |
|
83 |
In my experience having hardware raid cards has allowed me to get more |
84 |
per web, db, app, or mail server than software RAID resulting in more |
85 |
work for less money even with $ up charge for the raid card. |
86 |
|
87 |
I'd be willing to use the spare Dell 2850 with five drives and a raid |
88 |
card to do some testing this weekend, but someone will need to offer me |
89 |
a free sushi dinner or two if I'm right. Any takers? |
90 |
|
91 |
kashani |
92 |
-- |
93 |
gentoo-server@g.o mailing list |