Multithreaded Musings
Stand back - I'm a scientist!
Archives
« November 2009
MonTueWedThuFriSatSun
      
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
      
Today
Click me to subscribe
Search

About Me
Known throughout Sun as a man of infinite wit, of jovial attitude, and of making things up about himself at the slightest whim.
Links
 

Today's Page Hits: 2

Main | Next page »
Tuesday Jul 07, 2009
A ZFS Taxonomy
Nevada
Build
Version
Introduced
Features
Added
36 1 ZFS introduced*
38 2 Ditto blocks for meta-data
42 3 Hot spares
Double-parity raidz
62 4 ZFS command history
5 Added gzip compression
6 ZFS boot
68 7 Added slogs
69 8 Delegation
77 9 CIFS support
Filesystem-specific quotas
78 10 L2-arc
94 11 Improved scrub/resilver performance
96 12 Snapshot user properties
98 13 New "snapused" readonly property
103 14 Added passthrough-x to aclinherit property
114 15 User and group quotas
116 16 STMF property for COMSTAR
* Note: while ZFS was nominally introduced in build 27, the on-disk format was still undergoing changes until build 36, despite the fact that they were all labelled "version 1".
Posted at 02:28PM Jul 07, 2009 by Mark Musante in ZFS  | 

Thursday Mar 26, 2009
Cloning Humans: Bad; Cloning Filesystems: Good

Imagine you could snapshot yourself and generate a clone from that snapshot. On the surface, that would seem like a cool idea, but scratch the surface and you quickly run into problems. These problems could be ethical (what rights does your clone have to your own possessions?), or they could be practical (how can the planet sustain the population explosion?). Fortunately, with zfs datasets, snapshotting and cloning not only are permitted, they are actively encouraged.

Over at the Xerces Tech site, a recent article outlines how to use zfs clones in order to safely do an apt-get on their Nexenta box. Take a look at the article called Unbreakable upgrades, ZFS and apt-get to see how it's done.

Posted at 05:00PM Mar 26, 2009 by Mark Musante in ZFS  | 

Sunday Feb 15, 2009
ZFS Mandates
While making ZFS required for court and police NAS/SAN devices it would be nice if people would use ZFS because it is so obviously better, not because its use is mandated.
Posted at 06:18PM Feb 15, 2009 by Mark Musante in ZFS  | 

Wednesday Jan 28, 2009
Building a ZFS server

Have you ever wanted to try building your own ZFS-based file server? Jermaine Maree has done just that, and blogged about it. Start with part 1.

Posted at 03:02PM Jan 28, 2009 by Mark Musante in ZFS  |  Comments[1]

Thursday Jan 01, 2009
Don't shout at your JBODs
They don't like it!
Posted at 07:14PM Jan 01, 2009 by Mark Musante in Storage  | 

Tuesday Dec 23, 2008
Environmental Disaster
A TVA ash pond has flooded the Tennessee River valley. This is a huge disaster, but I didn't hear about it through the normal news channels. Why not?
Posted at 08:38PM Dec 23, 2008 by Mark Musante in Personal  |  Comments[1]

Wednesday Nov 19, 2008
ZFS can boost performance

Even a suboptimal configuration can result in a performance boost. The most interesting thing, I think, is the ease with which the zpool was created.

I wonder what kind of performance numbers this user would see with the 7110 compared with the Dell Powervault. The 7110 can hold 14 146gb sas drives, whereas the Dell uses 14 146gb scsi drives, so comparing power utilization would be interesting as well.

Posted at 03:24AM Nov 19, 2008 by Mark Musante in ZFS  | 

Thursday Oct 23, 2008
ZFS User Directories on OS X
Check out this blog post on setting up OS X with zfs-based user directories.
Posted at 02:58AM Oct 23, 2008 by Mark Musante in ZFS  | 

Tuesday Oct 07, 2008
Why it's going to take me forever...

One of the things I'm trying to do, besides have a full time job at Sun and raise three kids (another full-time job, even with my wife's help), is to learn Japanese. I'm not in any great rush, so I make time where I can and I try to spend at least a few minutes a day on it. I've started by trying to learn the kana writing system before I move onto the more complex kanji. There are two primary kinds of kana, hiragana and katakana. The former is used when writing Japanese, and the latter is used when writing "foreign" words, including words that have become part of the Japanese language but were borrowed from, for example, English.

Sun, as you're probably aware, encourages blogging from its employees around the globe, and more than a few of these are from Japan. Here's an example of one: キャンパス アンバサダ (I hope that came through).

Here's how the entry looks in my browser - I'll include it here in case your browser doesn't show the characters correctly

What I've been doing, in order to practice my kana, is to pick out the katakana symbols from the entries and see if I can work out what word it is in English. Here's the title of the entry I linked to, broken down kana by kana:

KanaEnglish pronunciation
first word
キャkya
n or m
pa
su
second word
a
n or m
ba
sa
da

Spelling it phonetically, we get kyanpasu or kyampasu for the first word, and anbasada or ambasada for the second. And I was puzzled. Contrast that with another katakana word that appears in the blog entry: "オリエンテーション", which is o-ri-e-n-teh-sho-n ... orientehshon ... orientation. Obvious, right? What's kyanpasu? Words that end with ス tend to have the final 'oo' sound dropped, so it becomes kyanpas when saying it aloud. I give up, so Google Translate tells me: Campus. This is where I hit my head on the desk. Why does Campus start with キャ? The a in kya sounds like the a in father, not the a in campus, so maybe that's why kya is used to differentiate it from the カ character ('ka') which also sounds like the a in father?

Of course, once I saw the first word was 'campus', it's easy to figure out the second word is Ambassador. Which I would have spelled, apparently incorrectly, アンバサドル

Posted at 03:01PM Oct 07, 2008 by Mark Musante in Hobbies  | 

Tuesday Aug 19, 2008
Cache on hand

When we think of a cache, we think of a way of storing information "closer" to the place it's needed. Most general-purpose CPUs, for example, have an on-board cache which is used to avoid accessing main memory - after all, the memory that's on the same die as the CPU is going to be quicker to access than the RAM chips. Filesystems use caches of RAM to make disk accesses appear to be quicker, as RAM chips are much faster than moving a mechanical arm across a spinning disks. If we're lucky, and if we've got a good caching algorithm, we can get an impressive speed boost by keeping the right bits in RAM. Likewise, CPUs get a speed boost by keeping the right instructions and data on chip.

Caching is not limited to CPUs and filesystems, of course. Most browsers maintain a cache of pages, of images, of css files, of javascript, and of any other bit of information that is useful for displaying web pages. By using a local on-disk cache (some of which is going to be in RAM anyway, thanks to the filesystem), browsing appears to be much quicker than it would if the browser had to re-load every single image from a distant web site. The browser does check to see if any files need to be retrieved again (see http's 304 Not Modified response code), so there is some over-the-wire activity, but that's about it.

All of this is a long-winded way of saying I was amused by the second bullet item here (from Apple Insider):

Either the cache is poorly implemented, or the users reporting this information are confused.

Posted at 12:56PM Aug 19, 2008 by Mark Musante in Algorithms  | 

Wednesday Jul 16, 2008
Good n' plenty

This is a great example of doing well AND doing good. Proof positive they're not mutually exclusive.

Posted at 03:46PM Jul 16, 2008 by Mark Musante in Sun  | 

Monday Jun 16, 2008
ZFS In The Wild, Part 5

It's been over a year since I last posted sightings of ZFS around the web, so it's high time I offered another list.

Posted at 07:59PM Jun 16, 2008 by Mark Musante in ZFS  | 

Friday Jun 13, 2008
Time Flies

It was nearly a year ago that I first made this screenshot:

Since then, I have done quite a number of different things, all related in some way to getting zfs to install and boot. Some of these things also involved teaching Live Upgrade to understand zfs datsets.

But now I'm starting to see that screenshot elsewhere, virtually unchanged from that fateful day long ago when I used the original to help design the changes needed for the text based installer.

Here are some: The Sect of Rama | Number 9 | Otmanix' Blog | Osamu Sayama's Weblog

It's really exciting to see it get out there and for it to be used outside of the development and test teams. Of course, there are some CRs being filed, and there are some things we'll need to address, but it's great nevertheless.

Posted at 04:20AM Jun 13, 2008 by Mark Musante in Bloggers  |  Comments[1]

Wednesday Jan 02, 2008
That Seattle Fireworks thing...

It's too bad they weren't using a filesystem that could detect and automatically repair corruption. Ah well.

Posted at 05:44PM Jan 02, 2008 by Mark Musante in Bloggers  | 

Friday Dec 21, 2007
It's easier than you thought

To all Windows-only admins:

Back in the olden days, I used DOS. I kinda got used to how scripting worked, and wrote many a .BAT file. Then I was shown Unix (an early System V version) and my eyes were opened. I understood what a real scripting language could do.

On the ZFS mailing list, I saw a message from someone who underwent a similar experience.

ONE OF US

One of us

One of us!

Posted at 01:07PM Dec 21, 2007 by Mark Musante in Coding  |