qatsi: (wally)
[personal profile] qatsi
Book Review: Elasticsearch in Action, by Radu Gheorghe, Matthew Lee Hinman, and Roy Russo
In a previous job I'd come across Solr, but these days, I'm told, Elasticsearch is where it's at, at least where I work now. Ultimately, both are built on top of Lucene; which does, actually, make you wonder whether that has any competitors.

This book was published in 2016, but one of its weaknesses is that it's very clearly somewhat out-of-date already, as it refers to the "latest" Elasticsearch being 1.5, with 2.0 in development. (There's been some creative arithmetic, but the current release is 5.x, and over that time there have been significant changes to some APIs.) Uncharacteristically for Manning, I did encounter a few badly worded sections, or examples or diagrams that didn't make sense or had incorrect values; cumulatively these things make me wonder whether the book suffered long delays and then finally got pushed out in a hurry without adequate proofing.

Hopefully the underlying principles remain much the same, with chapters on basic CRUD operations, index and analyzer configuration, search syntax, and relevance scoring. It is always against my intuition that many search queries are exact, when my expectation is for the results to have some degree of fuzziness; of course, in practice that's where the scoring comes in. There are also chapters on techniques for modelling parent-child relationships between documents, scaling, performance, and administrative tasks. Somewhat unfairly, several topics are relegated to appendices, including geo-search, hit highlighting, percolation (which was completely new to me but seems potentially very powerful, discovering which queries will match a given document) and completion/suggestion. It seems to me these are at least as, if not more important, than parent-child relationships and aggregation functions, but maybe that's just my use case.
Anonymous( )Anonymous This account has disabled anonymous posting.
OpenID( )OpenID You can comment on this post while signed in with an account from many other sites, once you have confirmed your email address. Sign in using OpenID.
Account name:
If you don't have an account you can create one now.
HTML doesn't work in the subject.


Notice: This account is set to log the IP addresses of everyone who comments.
Links will be displayed as unclickable URLs to help prevent spam.


qatsi: (Default)

September 2017

34 56789
17 1819 202122 23

Most Popular Tags

Style Credit

Expand Cut Tags

No cut tags