Whatever happened to the Google Search Appliance (concept)?

Yaoshi

Ars Scholae Palatinae
793
In every single company I have worked for a constant blocker/source of pain/waste of time has been actually finding relevant info on whatever I’m supposed to do, when said info is internal to the company and not on the public internet. I would be willing to bet the amount of time and productivity lost because of this would be a significant chunk of the world’s GDP.

This is the kind of info that could be in any of confluence/sharepoint/the recesses of whatever (usually godawful) intranet, other wiki or wiki-like system or - shudders - some Slack thread from months ago only to be surfaced by an alien civilization surveying the crumbled ruins of our society (I joke, but only slightly - in my experience Slack’s search is utter shit). So in the end the only recourse is to know someone who can at least point you to the general direction of what you need and hope for the best.

I remember reading about the Google Search Appliance in the distant past, and always thought it was a killer concept: a Google search field - when Google actually found things - for an org’s internal documents and knowledge, without the risk of having the data leaving the company’s network. Afaik the thing actually crawled through everything accessible in the intranet and did its page rank sorting thing, just like when Google was good.

Yet the Appliance has long been discontinued and nothing else seems to have come to do the same thing.

Why?
 

steelghost

Ars Tribunus Angusticlavius
6,245
Subscriptor++
We are a Google Workplace shop at work, and part of that is Google Cloud Search, which effectively does what you describe, but across Google Docs, Gmail, Chat, etc etc. for our "Workplace".

It's pretty good, but of course encourages you going all on their ecosystem, which I suspect is one of many reasons the appliance didn't pan out.
 

KD5MDK

Ars Legatus Legionis
23,215
Subscriptor++
Another challenge is probably how do you harmonize access controls for search result summaries across internal users if the search appliance is vendor agnostic? You can show a link that gets a 403 when the user follows it, but usually search results include a summary or excerpt of the content location and revealing the existence of a file can be sensitive enough. Even if authentication is always via SSO the search appliance needs to be able to understand the authorization criteria for each data source. Or just collect the results from the index and query each system to see if user X has access to that document right now.
 

w00key

Ars Tribunus Angusticlavius
9,134
Subscriptor
So yeah, you need ACL on the search index level too. That's how they get you to sign up for the expensive enterprise level. For example, this one does search/LLM with ACL:

https://cloud.google.com/products/agentspace?hl=en


Plebs and SME just have to drop sources they have access to into LM notebook manually, and know not to share them outside of people who are allowed to know it, especially a challenge for finance / HR documents.
 
  • Like
Reactions: sryan2k1