Google Mini 2.0

Inside the blue Google box is another box, also printed with the Google logo. Apparently Google have too much money to spend. In our European model, this box contains all the cables we'd ever need, plus a scary installation manual which was chose to ignore, other than to marvel at how many pieces of cardboard it was shrink-wrapped to for no apparent reason.

For some reason, Google provided a whole variety of cables to match a very international selection of power sockets (CEE 7/7 for most European countries, BS 1363 for the United Kingdom, CEI 23-16/VII for Italy, SEV 1011 for Switzerland, NEMA 5-15 for North America), but completely neglected to provide an IEC C13 to C14 cable, even though they insist the device must be connected to a UPS.

In the end, we didn't use any of the provided power cables, but we certainly appreciated the given network cables. The yellow cable is a nice long standard Ethernet cable, which means Google understand long cables are necessary. The orange cable is a cross-over cable for initial configuration of the box, and even comes with a label on it which tells you the IP address of the box to use. No manual required: Truly geek friendly!

The Google Mini itself has a funky, but obviously expected, Google design, and looks pretty nice. The box is disappointing for lovers of blinking lights, though, since the only light on the front of the box is next to the only button on the box, and both are related to power. Yes, I was disappointed.

The box itself is really just a PC in a swish box. That's pretty normal, considering it's the same thing for most so-called appliances these days, although most cover this fact up more than others. The idea follows the typical turn-key style network appliance. Since they hadn't covered it up, we freed up a port on our KVM switch to take a peek.

Perhaps in an effort to try to make those of us disappointed in the lack of blinking lights, you may notice something else hidden in the box. Google have provided an IT Manager sized extra large tee-shirt to help win us back after that whole blinking light debacle.

With no space free anywhere else, we had to mount our Google Mini in a two-pole rack, which has worked out fine since the box isn't too heavy. Connecting the cables was simple, since you initially only need the orange cable for configuration and the power cable. We also connected everything else so that we could have a look at the boot-up sequence.

As can be predicted, the Google Mini runs Linux, customised for the job. We never expected that the base would be RedHat; rather we thought Google would have put together their own system.

The Google Mini took a very long time to boot, however from watching the output on the console we noticed it had a lot of data to build. When the box was ready to go, it played a little tune, and prompted us to log in on the console. We were also able to use the administrative network interface (LAN2) to configure the box using a web browser. To help, this port has a DHCP server enabled to assist configuring your laptop, so you don't even need to guess the subnet configuration, even though it's pretty obvious.

The box also suffers an extremely long shut-down time which is especially annoying since this required before unplugging the box. Little consideration is given for automatically shutting down the Google Mini during a power failure, and since the box takes us around 10-15 minutes to shut down it's a little ridiculous there is no support for this.

After it was all set up, and a quick look at the manual, we signed up for a Google Mini support account. Everything looks good including the software update (our box already has the latest software from August, version 4.4.102.M.36). The fantastic thing about the software updater is that you can test the software on the built in test site and then accept or revert the update if it doesn't work well. The live side of the box that end-users are exposed to is only updated once you accept the upgrade. This fits into our test cycles perfectly, and Google deserves credit for having some forethought.

The administration console is fairly cut down and simple, although I would personally recommend people who want to get the most out of the Google Mini rush off and learn regular expressions (regex), since it will make your life easier. The console itself contains configuration for the box, as well as the crawler, and some statistical and logging/reporting functionality. We've been running our box for around a week now, so our index already contains quite a few documents, but we're still playing with it.

Configuring the crawler is not as intuitive as it could be, but I suppose there's a reason to that. If you don't understand how crawling works, you'll need to read up a bit. You need to define your crawl in at least two steps. Firstly, you need to define a starting point for all crawls. This is where the box will initially go to start looking for documents to add to its index. From there, following URLs might be included.

The second step is to define which documents you will allow within the index based on regular expressions (or just a list of sites). The point to this is that the Google Mini comes with a limited number of document licenses (the base model is 50,000), and if you started crawling inside your network, you could very quickly start crawling outside the network too. Fear not, because the Google Mini naturally looks for robots.txt as well.

Google know that intranets inherently have password protection on both websites and SMB/CIFS shares, therefore they have provided a crawler access configuration page, where based on a regular expression, you can match a username, password, and NTLM domain to a site, or particular sites. This means you can secure an account with read-only access for the crawler to access your employee share drives, and intranet website. It's very easy.

Finally, the box provides reporting, logging, and statistics. Here you can dig into sites and find broken links, monitor your document license usage, find sites you need to cut down crawls for, and so on. The statistics aren't amazingly powerful, but they tell you what you need to know to tune the index as much as you need to.

For the accident prone, Google have also provided methods of rolling back indexes along with the obvious export/import configuration settings, etc. The Google Mini also comes with a completely flexible and customisable interface, allowing you to categorise groups or catalogues of documents together and even provide different layouts for each catalogue. The user pages for searching and returned results are completely configurable through XSLT, so you can go nuts. Google provide a basic interface for customising the XSLT files if you're squeamish. However, if you're really nerdy, you can even get search results as raw XML, and do what you want with them elsewhere. The box is flexible enough to be seamless, if that's what's required.

With plenty of third-party support on the internet with various customisations, the Google Mini is become an attractive search appliance with a low price tag, small foot-print and low maintenance. I'm now addicted to it, since finding company-wide information takes a couple of seconds now as opposed to several minutes. This may not seem like much, but in a sales driven company like ours, we believe our Google Mini will quickly become a major time-saver to the sales department, who are constantly looking for quick information, such as product specifications.

Unfortunately there are short-comings that are solved in the Google Mini's big brother, the Google Search Appliance, which is stackable, redundant, stored an amazing amount of extra data, can connect to corporate databases, supports search item plug-ins, per-user restrictions on search results, and more. The Google Mini pales in comparison to the Google Search Appliance, however the price difference is alarming. I personally hope that Google will add more features in newer Google Mini software releases.