They are my oldest and deadliest enemy. You cannot trust them.
If Hitler invaded Hell, I would give a favourable reference to the Devil.

Monday, February 25


Almost, But Not Quite

The Samsung Galaxy Note 8 ticks most of the boxes where the Nexus 7 is wanting:
  • Stylus
  • Rear 5MP camera
  • 2GB RAM
  • MicroSD slot (up to 64GB)
There's a couple of things that could be better.  The CPU is a quad-core 1.6GHz A9, up from the quad 1.3GHz chip in the Nexus 7, but the new A15 core in the (Samsung-built) Nexus 10 is significantly faster.  And despite the slightly larger screen, the resolution is still 1280x800; I really want to see 1920x1200 at this screen size.  1280x800 is adequate, but a little short of ideal.  I've heard that the screen resolution is limited by the Wacom digitiser - the Note 10.1 is 1280x800 as well, where the Nexus 10 is precisely double that, 2560x1600.

It sounds great for my tablet use (reading, games, and note-taking), but we'll have to wait and see on price.

Posted by: Pixy Misa at 09:43 AM | Comments (3) | Add Comment | Trackbacks (Suck)
Post contains 149 words, total size 1 kb.

Friday, February 22



So exactly how long has my network switch been subtly broken?

I've been a bit down on the cheap LaCie NASes all this time because they just don't deliver the claimed performance.  At least, not on RAID-5.

LaCie claim 70MB/s.  I only ever got around 20MB/s.

Until I swapped out my network switch just now.  Bang.  70MB/s instantly.  On RAID-5.  The NAS was never the problem.

The new switch is the exact same model as the old one, except, apparently, that it's not horribly broken.

Now I'm just a bit down on the cheap LaCie NASes because the two new ones I bought (without disks) just plain don't work.  They power up and get an IP address (I can see them on the DHCP server), and I can ping them, but they don't do anything else.  You can't even turn them off - the power switch has no effect.  Behaviour is identical on both devices, whether they're configured with just one disk, a full five, or none at all.

Posted by: Pixy Misa at 04:25 PM | Comments (7) | Add Comment | Trackbacks (Suck)
Post contains 170 words, total size 1 kb.

Wednesday, February 20


Akismet Deployed

Just deployed Akismet as a second-source spam filter for Minx.  It only needed a little tweaking and is now running happily.

If you get blocked - you shouldn't, but if you do - please email me.

Update: In unrelated news, I accidentally deleted all my comments.  That's the comments I'd posted myself, not the ones on this blog.  I've put 99.5% of them back where they were; the remaining 9 are in the system but floating about detached from the posts.  I'll fix those later.

Posted by: Pixy Misa at 07:04 PM | Comments (6) | Add Comment | Trackbacks (Suck)
Post contains 87 words, total size 1 kb.

Monday, February 18


Oh Look, A Brick

My Nexus 7 is not a happy bunny.  It won't even boot in recovery mode.

Also, either my network switch, or my LaCie NAS - or both - are playing up.


Update: Found the super-double-extra-secret Android factory reset menu, and Android factory resetted it.  It's working again!  Now all I need to do is reload 300 apps and 600 books.  Interestingly, it reset to Jellybean 4.2.1, not the 4.1.0 that it originally shipped with.

My saved games are gone, but I wasn't in the middle of anything, so no great drama there.  Well, my capital city in Triple Town is toast, but okay.  I still need to see if the Kindle app restores the magazines I'd downloaded or if they've expired; I was wondering what would happen if I got a new tablet, and now I get to find out without even having to spend $299 first.

Update Twi: Yep, magazines are good, but it looks like it wants me to download all 600 books again one by bleedin' one.  The Android Kindle app leaves a certain amount to be desired.

Posted by: Pixy Misa at 10:48 PM | Comments (9) | Add Comment | Trackbacks (Suck)
Post contains 186 words, total size 1 kb.

Friday, February 08



In other database-geeky news, MongoDB 2.4 has hit RC (release candidate), with 2.4.0-rc0 quietly sneaking onto the download site.

This is the first version of MongoDB to support full-text indexing.  That's currently labelled experimental; it will be interesting to see how well it works and if it's ready for not-quite-enterprisey use, say, to pick an example entirely at random, in a multi-user blogging system.

Posted by: Pixy Misa at 03:01 PM | No Comments | Add Comment | Trackbacks (Suck)
Post contains 65 words, total size 1 kb.

Thursday, February 07


No Going Back

Intel's DC S3700 SSDs - the followup to the reliable and affordable but otherwise unremarkable 710 series - are now shipping.

Why they couldn't just call them the 720 or 750 or something I don't know, but if you work with enterprise databases at all, you want these drives.

Random read performance increases over the 710 series by 2x.
Random write performance increases over the 710 series by up to 15x.

The S3700 also comes with a consistency guarantee - with a queue depth of 32, 99.9999% of writes will take less than 20ms.  (At an average of 36,000 IOPS, using 4KB random writes over the entire device capacity.)  If you look at load graphs of older drives under heavy load, you see very noticeable drops in performance at irregular intervals as they run out of free blocks and have to reclaim space.  This just doesn't happen with the S3700.

It's also 40% cheaper then the previous model.

If you're still running databases on spinning disk, it's past time to move on.

Posted by: Pixy Misa at 06:40 PM | Comments (1) | Add Comment | Trackbacks (Suck)
Post contains 175 words, total size 1 kb.

Wednesday, February 06


MeeSQL 5.6

MySQL 5.6 is out today.  This one has enough enhancements to be worth a 6.0 version number, though hopefully without the usual .0 bugs.  In particular, it's the first version of MySQL to ship with full-text index support for InnoDB, previously only supported in MyISAM.  It also supports live table migration for InnoDB, another thing you could only really do with MyISAM before.

The problem with MyISAM is that its concurrency support is miserable.  It doesn't have real transactions, and while it can support as many simultaneous reads as you like, writes lock the entire table.

The problem with InnoDB, at least, if you look back five years or so, is that it is significantly slower and larger than MyISAM.  These days, though, neither of those is true any longer.

I tested copying 100,000 posts from Minx into nearly-empty tables built with MyISAM and both compressed and uncompressed InnoDB.  I say "nearly" empty because MyISAM has an optimisation trick for loading data into empty tables, and benchmarking it that way doesn't give you realistic performance figures.
  • MyISAM: 75.63 seconds / 64M data + 74M indexes = 138M
  • InnoDB Uncompressed: 34.6 seconds / 152M
  • InnoDB Compressed (8K block size): 49.41 seconds / 76M
So InnoDB uncompressed is 10% larger than MyISAM but over twice as fast, and InnoDB compressed is 40% smaller than MyISAM and 50% faster.  This is on a table with 63 fields, ten regular indexes, and three separate full-text indexes, so it should be exercising the storage engine pretty thoroughly.  More write-optimised tables would of course be faster to load.

At that rate, it only takes about an hour to load the whole post and comment history into InnoDB - so that's exactly what I'm planning to do.

Those benchmarks above were actually timed while that full load was running, which would likely slow InnoDB down a bit; I'll re-run them once the load finishes.  But it's not an unreasonable thing to do, since one of the major points of InnoDB is that it can run multiple write threads simultaneously.

Posted by: Pixy Misa at 09:02 PM | No Comments | Add Comment | Trackbacks (Suck)
Post contains 345 words, total size 2 kb.

<< Page 1 of 1 >>
68kb generated in CPU 0.0291, elapsed 0.1744 seconds.
55 queries taking 0.155 seconds, 367 records returned.
Powered by Minx 1.1.6c-pink.
Using http / / 365