AnandTech Search goes Google

by Jason Clark on 9/5/2005 9:41 PM EST
Comments Locked

48 Comments

Back to Article

  • bellwether - Thursday, November 29, 2007 - link

    This is a great starting point for search for small businesses. Google's algorithm is effective, but the problem is that the result page sends the user to the Google Mini itself (so they leave your website), and it is in Google's format. XSLT is supposed to help you modify this, but doesn't do that good of a job.

    This http://www.components4asp.net/GoogleMini/">custom google mini website search page has something for ASP.NET that lets you add in image thumbnails to the search result and integrate the search into a regular ASPX page that's part of your website. Plus, there's a 30 day free trial. Definitely worth taking a look at.
  • fzkl - Saturday, September 10, 2005 - link

    The dell memory is probably used because it has life time warranty.
  • mini - Friday, September 9, 2005 - link

    What is the OS used in the Google Mini?
    Could you please post more administration snapshots?
    Tks
  • Eirikur - Friday, September 9, 2005 - link

    I suspect some of your problems with the Full Text Search feature of SQL server might be related to how it breaks text into words and sentences. The word breaker will break by punctuation which is horrendous when it comes to version numbers. The word breaker will look at a version number like "2.0" and decide that "2" and "0" are two separate words in different sentences. Then it will throw both away since it ignores single letter words. In a version number like "2.82.1" only "82" will get indexed.
  • jberry - Wednesday, September 7, 2005 - link

    Does anyone know how the Google mini counts the 100K page limit with dynamic websites??
  • fishy - Wednesday, September 7, 2005 - link


    So...When are going to overclock this thing?

    ok, just k/d....
  • PassMark - Tuesday, September 6, 2005 - link

    There are much cheaper solutions around that you can run on your existing hardware and have similar performance without a limit of 100,000 pages.
    e.g.
    The http://www.wrensoft.com/zoom/">Zoom Search Engine for $99
    http://www.wrensoft.com/zoom/">http://www.wrensoft.com/zoom/
  • Brickster - Tuesday, September 6, 2005 - link

    I imagine there are certain documents that you would want only certain users to have access to. How do you control access to the documents that Google has indexed? Does it just return everything despite and document-specific, access security policy?
  • Verdant - Tuesday, September 6, 2005 - link

    hence why the pricetag on the big brother is next to useless imho...

    i can't see anyone using anything besides the mini for indexing something like a website, for knowledgebases and the like you need a lot more than just a way to search.
  • Brickster - Wednesday, September 7, 2005 - link

    Dude, you should see our company. A full featured search engine alone based on Google would do wonders for our cess pool of organization that is our intranet and file servers. For some, that is enough.
  • zmagaw - Tuesday, September 6, 2005 - link

    there are a few methods - including creating separate collections by user type filtering out or in urls by pattern matching
  • Brickster - Wednesday, September 7, 2005 - link

    Actually, I found you can achieve that with an upgrade... of course :)

    http://www.google.com/enterprise/feature_compariso...">http://www.google.com/enterprise/feature_compariso...

    Here is the Appliance feature:
    Secure Content API - Search across secure content using Google's Authorization API to integrate into existing access control systems.

    Looks like the Mini doesn't support secured content.
  • zmagaw - Tuesday, September 6, 2005 - link

    we signed a non-disclosure that said we couldnt open the google search appliances... although the hardware looks simple and run of the mill... the software is not... a lot of open source stuff on that puppy but execution is everything... the support we got though was horrible... 2 day respose times... so not easy because the software is full of bugs that are not easily diagosed... hardware failures - disks... and speedy google working with large corporations has been seen as a daunting task for the bright people at Google
  • nadirshakur - Tuesday, September 6, 2005 - link

    What is the warranty on these puppies. Hey didn't Anandtech void there's by opening it like that and showing the whole world they did.
  • flatblastard - Tuesday, September 6, 2005 - link

    Thats okay, if the RAM/CPU goes bad, I'll sell them my old p3 450Mhz system I got laying around for spare parts. Heck, I'll even give them a sweet deal.....$1999.95 and I'll even throw in Windows 98 (not SE)..... ;)
  • deathwalker - Tuesday, September 6, 2005 - link

    Is this really important when it comes to your experience when visiting the AnandTech website? I guess I'll get blasted for that statement!! So much really good stuff that could be the news of the day...this article is just cannon foder...something to fill the need for a new article to read on this day.
  • Jason Clark - Tuesday, September 6, 2005 - link

    Guys, we don't crawl every day :) It crawls 3-4 times a week, since large articles are on the front page, searching for them is pretty unnecessary.
  • Gooberslot - Tuesday, September 6, 2005 - link

    Is it normal for servers to give you no control like that? I wouldn't want anything that had a bios password that I couldn't change.

    I'm also surprised that you can even get a P3 or a P3 motherboard anymore.
  • zmagaw - Tuesday, September 6, 2005 - link

    i think when you are buying appliances yes... the reason google does this... you would be able to decompile their software on that hard-drive which is formatted in a google HD format - or so I have heard
  • smn198 - Tuesday, September 6, 2005 - link

    You could remove BIOS password but then you loose all the BIOS info as well and maybe they are doind something special there.
  • Calin - Tuesday, September 6, 2005 - link

    Pentium III processors are still offered in those 1U servers. The reason would be (probably) cheap price for good performance and lower thermal load than any other competitive Intel processors.
    Low thermal load helps a lot for dual processors servers.
  • flatblastard - Tuesday, September 6, 2005 - link

    Not only a p3 mobo, but PC133 ram labeled for DELL!?!? I guess Google is buying up all the old junk and putting it to good use.
  • bhtooefr - Tuesday, September 6, 2005 - link

    Heck, I've got a Dell PowerEdge 350 (1U, single 850 P3, i440BX chipset) sitting in front of me, and the RAM's not even labelled Dell...
  • flatblastard - Tuesday, September 6, 2005 - link

    making a fortune in the process....
  • brownba - Tuesday, September 6, 2005 - link

    Ok, so I tested it with this query:
    google mini search server
    - it came back with 18700 useless results.
    I also tried the title of the article:
    anandtech search goes google
    - 712 useless results

    how long does it take to crawl?
  • glennpratt - Tuesday, September 6, 2005 - link

    I'm guessing jason clark meant to reply to you.
  • Rock Hydra - Tuesday, September 6, 2005 - link

    I like it. I tested it out and got the returns I was expecting. Very Google-y style. Now if they implemented something this well into the forums search....maybe another day.
  • TheInvincibleMustard - Tuesday, September 6, 2005 - link

    So, naturally, I searched for iram ... returned zero results, but it did suggest i-ram as a possibility. So I clicked that link ...
    doh ... "I" is a very common word and so was not included in my search, meaning all I actually wound up searching for is "RAM" (of which there was several thousand entires, and not one of the top few was actually about the I-RAM product), so perhaps a bit more tweaking is in order ;-)

    Granted, though, the search did only take 0.02 seconds! :-D
  • dvinnen - Tuesday, September 6, 2005 - link

    I was playing around with it to. Did the i-ram search also, the first artical presented was an artical from 1997 about memory terms (think EDO). The actuall i-ram artical was actually the forth result presented. Hell, just a google search gives it as the 4th link in all the internets. Defently could use some tweaking (give added weight based on the date of the artical?) but looks to be a step up from the useless one built into SQL server you were useing.
  • glennpratt - Tuesday, September 6, 2005 - link

    Did you read my post?

    Did you click on the link that says 'i is a common word and was excluded' in the search? That would have given you a couple of choices on what to do to fix it.
  • glennpratt - Tuesday, September 6, 2005 - link

    Add quotes when you wan't a common word included, ie:

    http://search.anandtech.com/search?q=%22i-ram%22&a...">"i-ram"
  • Jeff7181 - Monday, September 5, 2005 - link

    I like the article... but more importantly, AnandTech Search isn't completely useless anymore.
  • Lifted - Monday, September 5, 2005 - link

    Hell, he will f---ing kill you!
  • Hi - Monday, September 5, 2005 - link

    i guess anand is dead



    :(
  • Verdant - Monday, September 5, 2005 - link

    at $3,000 it is a pretty good deal, but the 100,000 limitation is, well, a huge limitation, and the $30,000 pricetag on it's big brother is not that competitive with software solutions/crawlers.... especially since IT can only search 500,000 documents.
  • Googer - Tuesday, September 6, 2005 - link

    Are you kidding? It's a freaking P3 1.26GHz! I could Do much better for Two grand.
  • rajivdx - Friday, September 9, 2005 - link

    Make no mistake about it its bloody fast!! It makes the rest of the Anandtech site look *absolutely* slow by comparision.
  • Verdant - Tuesday, September 6, 2005 - link

    you've never looked at the prices of database software licenses, developers or content management packages before, have you?
  • Calin - Tuesday, September 6, 2005 - link

    There are two processors in fact - and you could buy something like that (no hardware) for $1,500 easily (maybe not so easily as the 1.26GHz Pentium III could be hard to find, but you could replace them with Athlons or Pentium4s at the same performance for cheaper).
  • JustAnAverageGuy - Tuesday, September 6, 2005 - link

    You're mostly paying for the software.
  • Jynx980 - Monday, September 5, 2005 - link

    Interesting read, I don't understand a lot of it, but interesting none the less. Were you nervous cracking it open? I'd be sweating like Roger Ebert if I was trying to crack open a $3000 pizza box.
  • Netopia - Monday, September 5, 2005 - link

    How long before we see a big brother for it caching the forums?

    Joe
  • PorBleemo - Monday, September 5, 2005 - link

    Waaaay too much information in the forums. You would max out the higher-end model easily.
  • Marlin1975 - Monday, September 5, 2005 - link

    Can you just not buy the software and install on your own hardware? Heck a 686b southbridge, does VIA even make those anymore, let alone SDRAM? That box must have had a lot of dust on it.
  • coomar - Monday, September 5, 2005 - link

    so searching on anandtech is going to be using google search engine

    i didn't know microsoft was responsible for the engine used right now
  • Verdant - Monday, September 5, 2005 - link

    i would hardly say microsoft was responsible... it was SQL server, but you can hardly say it is a search engine in a box
  • bellwether - Thursday, November 29, 2007 - link

    Not that Google's search is much better (from a visual stand point). Heck, you had to have people develop http://www.components4asp.net/GoogleMini/">google mini add-ons so it could do things like show image previews and have a layout that doesn't look like a Google search.
  • leexgx - Friday, December 21, 2007 - link

    adding an comment to this is fun when last post was like 2 half years ago heh

Log in

Don't have an account? Sign up now