I was intrigued by Google’s recently announced DNS service. Honestly what caught my attention was the IP address they were using for the public service – 8.8.8.8. That has to be one of the easiest to type IP addresses in the IPv4 range. They haven’t triggered my flags about the end of the world surveillance paranoia that some people have every time Google sneezes too loud, and they aren’t playing games with NXDOMAIN records like OpenDNS does, which is nice.

I would recommend most casual users to switch over to it – however, I’m not planning on using Google’s DNS services at all. My local BIND servers are configured to cache DNS requests, so I get results back faster from my DNS servers (1ms, on average) than Google’s (10-20ms, on average). I may evaluate adding 8.8.8.8 as my upstream authority at some point, however.

I’ve been getting that itch recently to look over my backup system. Some things are working very well, and some… not so much. At the moment, I’m using Dropbox to sync about 25 GB of data between two computers at work and two computers at home. About a third of that is tar.gz’s of my home folder on both office machines, generated every Monday morning. The next 1/3rd is personal data I’d rather not lose, and the remaining 1/3rd is work documents. This makes it easy to work wherever I am. It also gets me 5 copies of the data for disaster recovery if you count the copy on the Dropbox servers themselves. It also functions as a backup, as you can restore point-in-time backups.

While Dropbox covers my working data (other than very large files), I need a backup for the OS’s. I don’t yet have a good one for my Solaris or Linux boxes, but for the Mac’s, I’m running Time Machine to local SATA drives. (I abandoned the Time Machine on Disk Images silliness a few weeks ago. It was far too slow when doing photo work, often taking 20 or 25 minutes to backup the changes. The host it was backing up to was memory and I/O bound – not a good mix.)

I also burn DVD-R’s every month of the tar.gz files created by my backup routines, as well as a tar.gz of the Dropbox folder itself, excluding the backups. Since most of what I do is text, it currently compresses down to about 8 GB. I include MD5 checksums of each backed up archive, to make sure I can verify my data later.

That covers all of my normal data. I also have a ZFS based file server which I copy all disk images, movies, music and other things I didn’t create (which I can re-rip from DVD or CD, or can re-download). I also push a disaster recovery copy of my Dropbox folder to the ZFS server, and use snapshots to manage the revisions.

Ironically, it’s the easiest stuff that isn’t working right. I mentioned awhile ago that Time Machine occasionally tries to back up my whole disk again. I’ve had to run in and beat it about the head quite a few times to get rid of teh stoopids. Looking at it now, there is over 130 GB of data that it’s backed up, incorrectly. My real concern is when it fills up the disk Time Machine will happily delete all the older backups up to and including the last mammoth backup. This defeats the purpose of Time Machine, which is to give you a nice “slice of time” backup. The other problem is that I use that drive as my scratch drive for Aperture and Photoshop work, as editing directly on the network isn’t quite fast enough. Once the drive fills up with Time Machine backups, I’ll have a real problem to deal with.

The next problem is that my alternate Solaris server is dreadfully low on memory (1 GB) and only has a 100bT NIC. It’s an older box, so memory isn’t cheap – a 2GB DIMM is $120, which is roughly twice the price of fully buffered ECC DIMMs is for my Mac Pro. I’m trying to find the right use for it – which at the moment is running a smaller 750 GB mirrored pool for photo backups. It’s slow, but it works. I’m honestly tempted to load XBMC on it and see if it can decode the large MKV’s I rip from my DVD’s. It would make the system a lot more useable, and if I can find a remote for it, it would pass the wife-friendly test.

What I’d really like to do is get a second internal drive for the Mac Pro for photo editing. That’s going to run me around $75. Next, I’d like to go ahead and get the SATA card and the drives to do the faster photo pool I mentioned in my Photo Management post. Of course, that is very much discretionary, and I’d probably prefer to get a second monitor first.

I was trying to comment on a ZDNet blog a few days ago, and I was forced to register for a ZDNet account. The bad part: they signed me up for a bunch of electronic newsletters without giving me an obvious opt-out. The only way to keep my ZDNet account and unsubscribe from the newsletters that I have found so far is to supply them with a lot of personal information (Name, Address, etc). I was worried about, until I noticed that GMail was auto-sorting the ZDNet newsletters as SPAM.

Nicely done, Google.

Today, my wife got me a Magic Mouse as a surprise gift. After using it for a little over an hour, I’m sold. It’s going to be hard to go back to work on Monday and use a ‘normal’ mouse again. Apple has made it fit in with the rest of the OS very well – scrolling is more natural with this mouse than it is with a scroll wheel, and if you’ve used a Might Mouse for any period of time, you’ll have no problem with the right click.

The wickedness is in the things it does that other mice don’t or can’t. For example, when using VLC, the single-side-swipe gesture fast forward or reverses. The up-down-scoll adjusts the volume.

I’m sold.

As I mentioned a while ago, I use and love splunk. I had bemoaned the fact that the Splunk 4 “Free” Edition wasn’t yet available. It is now. Go get it. I’ll wait. In 4, they have really upped the scale and speed of log ingestion, indexing and searching. My log server is a single core AMD Sempron box with 1 GB of RAM, and it’s wicked fast. Noticeably faster on the same hardware than 3 was. It’s not often a vendor ships faster products than previous iterations.

The irrelevance of News Corp in the media landscape, that is. Rupert Murdoch has announced that not only are they going to errect a paywall around their properties, they are going to remove themselves from the Google.com search indexes. Someone must not be looking at their server logs to see where their traffic comes from. I’d be willing to bet a significant portion on my next months’s salary on the fact that almost all of their traffic from external sources – from referrals, search engines and the like. Very little of it comes from another News Corp property, be it the Times or Newsday or the Wall Street Journal (or whatever else they own).

I’m also really curious to see if Microsoft tries to slip them some cash to get sponsored referrals – in effect paying for bing users to read the content. It’s something they could certainly fund, and would draw attention to Microsoft, who is also becoming increasingly irrelevant.

Twice today Time Machine has decided that it needs to completely backup everything on my boot drive that I have not specifically excluded. This is just under 60 GB of data, twice. I was shocked when I saw how much less space I had on the Time Machine drive, which prompted my inquiry. TimeTracker gave me a nice view of the changes, and what’s going on. I’m going to hold off disabling Time Machine until I see what things look like tomorrow morning.

I may have to get a second drive dedicated for Time Machine, but I’m going to be a little sad if I lose revisions of files because it decided to back my whole system up too often. All this will do is push me further into the ZFS camp. I’m still a little sad that Apple didn’t base Time Machine on ZFS – it would have been a much more elegant system for doing backups than what we have now. Of course, I’m even sadder that Apple dropped ZFS from Snow Leopard. Maybe one day…

UPDATE: everything has sorted itself out. The huge backups are slowly expiring, as they aren’t the last of the day, and the space is coming back. It’s still somewhat worrisome that it happened at all. I think the writing is on the wall – I need a dedicated Time Machine drive.