I really like everything on GSA expect the price tag. I am not much fan of Site Search since I have no control on Meta tag level. I wish they would have kept Google Mini as middle layer between GSA and SiteSearch where I was able to control search in custom <meta> tags. Perhaps also they could have put Mini to Virtual Image or cloud instead of having to buy hardware with it just to make it easier.
We use swiftype. Can't compare with GSS because never actually used GSS, but swiftype seems to have better APIs, responsive support staff and good looking UI out of the box for end users. And we are quite happy with it.
For an enterprise with mixed Windows and UNIX servers and SharePoint there better options than Google. For web based data, Google does great. For mixed media that has varying security constraints we found Google lacking.
An interesting anecdote about the Google Search Appliance, a bunch of companies complained about them in the beginning because they conflated "search intranet" with "search file shares" and of course Page Rank is not as effective on file shares as it is on web based systems.
2002 - "The first Google product for enterprises is released: the Google Search Appliance is a yellow box that businesses can plug into their computer network to enable search capabilities for their own documents."
We had one of these like a decade ago. I think it just sat in the DC collecting dust. Really wanted to look inside but they weld the damn thing shut and you have to ship it back to goog for maintenance.
Same, and I've seen that in several large enterprises since. Typically a case of infrastructure focused IT management saying "Google in _our_ DC, how cool would that be!", getting some boxes, then the inevitable "ok, now what?".
NOTE: Despite that scathing comment, I continue work in large enterprises... sure they have their downsides, but you can also work on really cool things if you seek them out.
They've historically (until this generation -- I haven't poked at our new ones) used commodity Dell servers, along the lines of a 2U PowerEdge. While they were welded shut, it was entirely possible to wipe them and reimage with your own Linux install, provided you had a way to boot from USB. We did this with all of the previous generations of appliances after they were replaced by the new contract. They became perfectly serviceable test boxes.
As far as I know your only options now are the GSA or Google Site Search. The mini is on Google's incredibly long list of discontinued products and services.
I've been doing a lot of research on enterprise search products lately and one of the biggest issues is security. Unleashing search on corporate systems where permissions aren't managed really well is a big risk.
[front matter: I worked on the Google Search Appliance and also administered Google's internal cluster of GSAs]
The GSA is actually pretty good when it comes to security.
There are a range of options (this was circa 2007, so the technology may have changed). You can configure the GSA to not show "snippets" in the search results. It's fairly easy to write a script that harvests an entire document from the index via snippets, so the GSA gives you the option to disable them.
At a higher level, even though it takes some amount of configuration to pull off, you can get a GSA to proxy a client's credentials to the top N search results, to determine whether or not the client would have access to that document.
If the client would not have access, then the result would not show up in the search results. Yes, this has a big performance impact, but it's the paranoid-level.
Again, this was how it worked in ~2007, so it may have changed since.
That has traditionally been the main objection to the google search appliance, or so I have been told - but the site now says the thing supports ACLs. Anyone has experience with that?
For some of our systems, we use Alfresco and/or Drupal, both which have Apache Solr as an easily-configurable search backend, and which works basically plug-and-play with the respective systems' content permissions.
Using the free versions of all three (Solr + Drupal + Alfresco), search is fast and access-controlled very easily.
I worked for a company that had a Google Search Appliance about 6 years ago, and, like others have said, it basically sat in a rack somewhere and collected dust once we started using some form of integrated search. The GSA was fast and relatively easy to use, but required an expensive initial purchase + ongoing licensing/update fees.
It is basically a self hosted scraper and search engine in one. You give it your URL and some display templates and it indexes your site in much the same way Google normally would. The only difference is that the results are stored and served locally. They charge based on documents indexed. I don't think it would be good for storing arbitrary data like Solr, just web pages.
Apart from the fact that this isn't a new product. From a ranking and relevance point of view, can the same (internet) search algorithms be reused for internal (intranet) search? Often permissions, links and usage are not reflective of the importance or reputation of a document like on the public web. Often emails carry a bulk of the decision making material.
Not to downplay the importance of an enterprise search offering, but I don't think Google just packages there standard search in a box nor do I think 1 single GSA box can solve enterprise search.
There are a few of these on eBay, I wonder what the pricing model is. Surely Google doesn't like them just popping up, but I guess if you're out of your agreement you can resell them?
It's interesting that this should pop up today. This morning I attended a Google-sponsored event where they spoke a little about GSA.
I'd have assumed that another HN reader was present but I'm in a city with a pretty non-existent tech community and the event only had ~30 attendees, the majority of whom were energy sector CEOs and GIS people.
38 comments
[ 3.0 ms ] story [ 99.8 ms ] threadhttp://www.coveo.com/en
For an enterprise with mixed Windows and UNIX servers and SharePoint there better options than Google. For web based data, Google does great. For mixed media that has varying security constraints we found Google lacking.
There is also a review of each one.
An interesting anecdote about the Google Search Appliance, a bunch of companies complained about them in the beginning because they conflated "search intranet" with "search file shares" and of course Page Rank is not as effective on file shares as it is on web based systems.
2002 - "The first Google product for enterprises is released: the Google Search Appliance is a yellow box that businesses can plug into their computer network to enable search capabilities for their own documents."
NOTE: Despite that scathing comment, I continue work in large enterprises... sure they have their downsides, but you can also work on really cool things if you seek them out.
The GSA is actually pretty good when it comes to security.
There are a range of options (this was circa 2007, so the technology may have changed). You can configure the GSA to not show "snippets" in the search results. It's fairly easy to write a script that harvests an entire document from the index via snippets, so the GSA gives you the option to disable them.
At a higher level, even though it takes some amount of configuration to pull off, you can get a GSA to proxy a client's credentials to the top N search results, to determine whether or not the client would have access to that document.
If the client would not have access, then the result would not show up in the search results. Yes, this has a big performance impact, but it's the paranoid-level.
Again, this was how it worked in ~2007, so it may have changed since.
Using the free versions of all three (Solr + Drupal + Alfresco), search is fast and access-controlled very easily.
I worked for a company that had a Google Search Appliance about 6 years ago, and, like others have said, it basically sat in a rack somewhere and collected dust once we started using some form of integrated search. The GSA was fast and relatively easy to use, but required an expensive initial purchase + ongoing licensing/update fees.
Not to downplay the importance of an enterprise search offering, but I don't think Google just packages there standard search in a box nor do I think 1 single GSA box can solve enterprise search.
I'd have assumed that another HN reader was present but I'm in a city with a pretty non-existent tech community and the event only had ~30 attendees, the majority of whom were energy sector CEOs and GIS people.