AWS Kendra: A Somewhat Elastic Approach to Enterprise Search

May 12, 2020

Elastic, Shay Banon’s Version 2 of Compass, has a hurdle to jump over. Elasticsearch has been a success. The Lucene-centric “system” which some call ELK has become a go-to solution for many developers. Like Lucidworks (It does?) and many other “enterprise search and more” vendors, Elasticsearch delivers information retrieval without the handcuffs of options like good old STAIRS III or Autonomy’s neuro-linguistic black box.

Amazon took notice and has effectively rolled out its own version of enterprise search based on … wait for it … the open source version of Elastic’s Elasticsearch. The service has been around since Amazon hired some of the Lucidworks (It does?) engineers more than five years ago after frustration with the revolving doors at that firm became too much even by Silicon Valley standards. Talk about tension. Yebo!

Amazon has reinvented Elasticsearch. The same process the Bezos bulldozer has used for other open source software has been in process for more than 60 months. Like the system’s Playboy bunny namesake, Kendra has a few beauty lines in her AWS exterior.

A tweak here (access to Amazon’s smart software) and a tweak there (Amazon AWS pricing methods), and the “new” product is ready for prime time, ready for a beauty contest against other contestants in the most beautiful IR system in the digital world.

Amazon Launches Cognitive Search Service Kendra in General Availability” reports:

Once configured through the AWS Console, Kendra leverages connectors to unify and index previously disparate sources of information (from file systems, websites, SharePoint, OneDrive, Salesforce, ServiceNow, Amazon Simple Storage Service, relational databases, and elsewhere).

Does this sound like federated search or the Palantir Gotham approach to content?

Well, yes.

The reason is that most enterprise search vendors like Coveo, Attivio, X1, IBM Omnifind (also built on Lucene), and dozens of other systems make the same claims.

The reality is that these systems do not have the bits and pieces available within a giant cloud platform with quite a few graduates of an Amazon AWS training program ready to plug in the AWS solution. For example, if a government agency wants the search in Palantir, no problem. Palantir deploys on AWS. But if that government agency wants to use Amazon’s policeware services and include search, there’s Kendra.

You can get a free copy of the DarkCyber Amazon policeware report’s executive summary by requesting the document at this link.

What does Amazon bring to the enterprise search party?

The company has more than 200 services, features, component, and modules on the shelf. Because enterprise search is not a “one size fits all”, the basic utility function has to fit into specific enterprise roles. For most enterprise search vendors, this need for user function customization is a deal breaker. Legal doesn’t want the same search that those clear minded home economics grads require in the marketing department. Microsoft SharePoint offers its version of “enterprise search” but paints over the cost of the Microsoft Certified professionals who have to make the search system work Fast. (Yep, that’s sort of an inside search joke.)

Amazon AWS provides the engine and the Fancy Dan components can be plugged in using the methods taught in the AWS “learn how to have a job for real” at a company your mom uses to shop during the pandemic. Amazon and Microsoft are on a collision course for the enterprise, and the Kendra thing is an important component.

The official roll out is capturing headlines, but the inclusion of Lucene-based search invites several observations:

  • Despite AWS’ pricing, an Amazon enterprise search system allows the modern information technology professional to get a good enough service with arguably fewer headaches than other options except maybe the SearchBlox solution
  • Enterprise search becomes what it has been for most organizations: A utility. Basic information retrieval is now an AWS component and that component can be enhanced with SageMaker, analytics, and other AWS services.
  • Amazon wins even if Kendra does not win the hearts and minds of IBM Omnifind, Inbenta, and Algolia users. Why? Most of the cloud based enterprise search vendors support the AWS platform. What are the choices? The wonky HP cloud? The “maybe we will kill it” Google Cloud? Azure, from the outfit that cannot update Windows 10 without killing user computers who activate game mode? Plus, dumping Kendra for another TV star inspired search system is easy. Chances are that, like Palantir, AWS hosts and supports that competitive system too.

Net net: The fight with Microsoft is escalating. The Bezos bulldozer will run over open source outfits and probably some AWS customers. But Kendra’s turning her gaze on the bountiful revenues of Microsoft in the enterprise. Will Amazon buy a vendor of Word, PowerPoint, and Excel clones?

Exciting times, maybe not just because of enterprise search? Why did those defectors from Lucidworks (It does?) embrace Lucene and not SOLR? Maybe they did that too?

Stephen E Arnold, May 12, 2020


Comments are closed.

  • Archives

  • Recent Posts

  • Meta