Unrealistic as it may seem, Griffin is real. She's in the Pajamachievements and does the comic over at RoosterTeeth.com

Rooster Teeth Shorts, Immersion (Pilot)

Not cool that Gizmodo didn't give them credit. These are the same guys that do the Red Vs Blue machinima.

Low cost hosting providers rarely guarantee backup and restoration services as part of the low cost package. It is often a separate item entirely that must be paid for in addition to the standard account. Not only this, many of the shared/virtual private server type providers do not offer any guaranteed recovery period if the server you happen to be on goes down. If you are experiencing an outage due to another user sharing your hardware being compromised and they take the server offline, often times the provider will do nothing to get your site back up and running quickly even if you have the data prepared to slam back onto a new system; You just end up having to wait. (First hand recent experience with a one-and-one vps: The hardware had a drive controller failure. We have full backups of the VPS via bacula and if they were willing to give us a second vps on a new server at the same IP, we could have slammed the data back onto the server and been back up within the hour. They instead made my customer wait 48 hours while they worked on trying to make that original server work.)

Regardless of who you pay for hosting, your data is your responsibility. Their backups are worthless if you never actually prove they are usable yourself. Plan for disaster ahead of time and you'll be better off. Plan it at several different levels: what happens when the data is corrupt? What do you do if the server catches on fire? What do you do if the city/region experiences a catastrophe? What do you do when Joe Constructionworker is installing sprinklers next door and puts their backhoe through the datalines feeding the center? If your provider is offering to cover any one of these with a solution (like paying them to backup the data for your restoration) find out how you get the databack and what kind of SLA's they have. If they back it up, but it's a 24 hour process just to get to the point where you can restore things, that may not work for you. Understanding your recovery process before you need to put it in place is one of the biggest failings of many users/companies offering web based service delivery.

Now, one of the more interesting lower cost providers I've run into lately is Linode. You have a bit more flexibility in dealing with scaling and failover and you can move your virtual private server to bigger and beefier hardware as your site grows. They are working on an inhouse backup solution, but realistically if you care enough about your data, you'll regularly backup offsite with scripts or your favorite backup program (bacula anyone?). Linode is targetted more towards those who can admin their systems themselves rather than needing pre-setup solutions with GUI's (not that you couldn't use something like Plesk yourself on it). You can slowly scale your system hardwarewise to machines that have less and less shared users and you can even use multiple virtual servers with virtual load balancers in front of it (they have some interesting support for having a private lan between your virtuals that keeps the traffic 'local' and won't count against your bandwidth usage. You could use multiple virtual nic's to do load balancing with LVS type setups if you wanted).

When it comes down to processor comparisons, I see very little compelling about this new AMD proc. The i7 920 is going to outperform it at most things, uses less power and is only 35 bucks more. Eventually for those of us always-on users, even the 10 watt savings of the i7 is going to kill the slight price advantage.

The only thing I see interesting here is the fact that you have more commodity boards to choose from, could do a slower upgrade (re-use your ddr2!) but this isn't any different than the currently line of quad proc amd chips, many of which can be had for cheaper and use less power.

Come on, AMD, you can do better.

LED lights are a cheap fun way to teach some basics. All you need is a battery (or even better, several different batteries with different voltages), an LED (or several LED's with different voltages), and a bunch of resistors.

You can get packs of green, red, and yellow LED's for less than 50 cents an LED. resistors are a buck for packs of 10. And batteries are batteries. Figuring out the resistor needed to light up an LED based on the voltage from a single battery or series of batteries can be neat.

If you want to take it a step further, bring in some 50 cent USB a-b cables. Slash them and toss out the B side, find the 5V and ground line, and have them figure out the resistor needed to light an LED for USB voltage (like a woot light!). USB power = 5V 100ma usually (it goes up to 500ma, but the driver usually has to negotiate it up; it should be 100ma; buy a cheap powered hub if you want to keep it safe from the computer). There are lots of links on how to figure out the voltage of an LED, this one is ok.

My favorite tools are a combination of the Trinity Rescue Kit linux boot cd and the Restorer tool.

It depends on the type of failure, but generally, I start with a ddrescue to get an image of the drive, especially if the drive is running bad sectors. Either I set the image to go to a secondary spare drive or I push it across the network. ddrescue is nice in that it doesn't bail when it hits those bad sectors, can run in reverse mode, and eventually it'll get as much as isn't corrupt on the drive into the image.

After establishing the image, the original failed drives go into ESD bags and aren't touched again unless they are to get shipped to one of the expensive clean room type places for their style recovery.

Most of the win32 drive recovery softwares out there can handle reading from an image file, so from here on out, I work with the images I took with ddrescue. Restorer has worked pretty well for me on getting things back from hard drives, CF cards, and even raid sets (figuring out the cluster sizes on the raid can be a pain if you don't happen to know them, but the software does support reassembling raid drives from the images you take of the single drives).

Most of the win32 packages out there have support for making the original images, but I haven't had as much luck with most of them when dealing with severely corrupted drives or with a large scattering of bad sectors. Either they take far too long to make it through the image or they end up failing to get by the bad sectors.

Regardless of what you end up picking, you don't want to use any of the recovery tools that advertise how they can fix the partition table and such on the drive, live . . . any recovery operation that thinks it is ok to 'fix' a drive with data on it you want to recover has the wrong mindset. The data is important, not making the drive work again.

Has anyone noticed any large maid-like robotic entities in orbit? More importantly to our future, were there any winnebagos with wings nearby?