Awhile back, I wrote a three-part blog series on federation. (Part One: “Make Federation a Part of Your Single Pane of Glass”) and why it should be part of your enterprise search strategy. Since then, we have noticed a huge shift away from a single-index strategy to a multi-index strategy.
In the first installment for this blog series, I am asking you to shed the old paradigm on enterprise search. As Steve Jobs would say, “Think Differently.” Strongly consider a multi-index strategy.
Many of our new and existing customers have taken this hybrid strategy of combining query-time and merge-time indexing to heart. Implementing this strategy has led to astonishing results including a better user experience, lower total cost of ownership, quicker time to deployment, and search results that are always fresh!
As I stated in my previous blog, enterprise search vendors historically positioned these two approaches as separate. They promoted merge-time indexing/single index as the superior strategy and even went as far as stating it is the only possible solution for enterprise search deployments. BA Insight disagreed with this thinking several years ago. We still do. In fact, we believe in this hybrid strategy even more vociferously.
Both work well with each other. It is not one versus the other.
The hybrid approach that is used for the internet is also the best approach for many enterprise search applications. Individual sites have their own search engines and often use index-time merge for search within their sites. But when you “Google” it, Google performs query-time merge and brings you spectacularly relevant results.
Pundits keep promoting decade old philosophies about why enterprise search is different. Their philosophies support the old way of doing things. This continues despite employee complaints about internal search not being even close to what they are looking for, which is something that resembles “Google- like” search.
Searching the web is the platinum standard for enterprise deployment of search applications, whether it is internal, customer facing, or on your website. With AI advancements and enterprise application vendors providing robust APIs, the hybrid search strategy is often the best approach.
Some “why do it the old way” example use cases for you to think about:
- If you use a modern, open search engine like Azure Cognitive Search, Amazon Open Search Service, or Elasticsearch, why crawl and merge O365/SharePoint into that index? Imagine the infrastructure costs for the new index (for potentially 100s of millions of items) and the risk with keeping it fresh.
- Subscription applications won’t allow indexing into another search engine. So, you don’t end up with a single pane of glass.
- In the legal market, the old way would have you index from a DMS like iManage and NetDocs into the vendor’s index. Not only is the infrastructure cost for 100 million+ items an issue, but keeping it fresh with the rate of change in legal is an even bigger challenge.
Read our related whitepaper on index-time and query-time merging for more perspectives on this topic.
In my next blog, I will provide details about a hybrid implementation at a visionary, biopharmaceutical company committed to discovering, developing, and delivering innovative medicine to patients.