Loading...
 
Skip to main content
17
Page actions
PDF Print Get a short URL Open in WebDAV

Unified Index Comparison

The Search and List from Unified Index has support for multiple engines. While all of them offer the same general functionality and connect to various functionality such as the content search, PluginList, PluginCustomSearch and various others, they will have different performance characteristics and some may offer additional features.

As a general rule, the engine can simply be switched and the index rebuilt without any additional change to the configuration.

The default settings are fine for all but the biggest Tiki projects (top 1%). If rebuilding your index takes less than 30 minutes, you generally have nothing to worry about.

There are so many parameters that we can't predict at how many pages, files or tracker items that you will hit an issue. It depends on many factors, including

  • Server resources (RAM, CPU, etc.)
  • Quantity of data
  • Type of data (tracker items vs wiki pages, etc.)
  • Number and type of tracker fields
  • Version of Tiki and how it's configured
  • Version of database and how it's configured
  • Etc.


So just stay with the defaults until/unless you hit a problem. Ex.:

  • A full text search becomes slow.
  • Error messages related to having too many fields


If you have or plan to have a big Tiki instance and you are concerned if Tiki will scale to your use case, you can use Faker to generate fake data on your server. You can also reach out to experienced folks.
For community support: https://gitter.im/tiki-org/community
For professional services: https://evoludata.com/

Overview

The unified index engines are:

  • MariaDB / MySQL / Percona Server for MySQL Full Text Search MyISAM
    • First version: Tiki12 Last version: Tiki27 (the default engine for all these versions)
    • Additional memory required
    • Fast indexing (can be 10 times faster than now removed Zend_Search_Lucene), slower/unstable query speed
    • No configuration required
    • Not customizable
    • Stored in tables in the database with a prefix of index_ and are stored in MyISAM even if the actual data is in InnoDB
  • MariaDB / MySQL / Percona Server for MySQL Full Text Search INNODB
    • introduced in Tiki28 (new default engine from now on)
    • No more limit on the number of columns (We now workaround INNODB limits)
    • Stored in tables in the database with a prefix of index_ and are stored in InnoDB. All the data tables should also be in InnoDB, but it's also possible the data is still in MyISAM. Ex.: an Tiki instance started when MyISAM was the default, and the conversion was never done.
  • Elasticsearch
    • introduced in Tiki12
    • Independent Java server(s), horizontally scalable
    • Feature-rich
    • Fast indexing, fast/stable query speed, decent/good results
    • Typically, Elasticsearch is set up as a cluster on different servers than Tiki (or using a third-party service), but it is also possible to install on the same server.
    • Customizable
  • Manticore Search


The system is designed for maintaining an autonomy vis à vis the engines. So more can be added later. No long-term data is stored in the indexes and it's fairly easy to switch from one to another. The next logical addition is OpenSearch. Please contact Marc Laporte if you have specific needs.

Limitations

MySQL/MariaDB MYISAM

  • This is used until Tiki27 LTS
  • Words with fewer than 3 or 4 characters will not be indexed unless the server configuration is modified. Variables name: ft_min_word_len and innodb_ft_min_token_size
  • Comes with an extensive list of English stop words, preventing many queries from working.
  • Can use a single index at a time. Depending on the query, performance can vary significantly.
  • Several limitations on the number of columns and indexes it can contain. Complex sites with many different query patterns may hit those limitations.
  • No support for field boosting, such as providing more relevance for hits on the title.
  • There is a limitation on the number of tracker fields. The limitation is quite high (2000+), but when you hit it, you need to move to another engine because MySQL/MariaDB has a hard limit. It is not possible to know in advance the precise number of maximum fields because some tracker field types require more than one column.
  • How to search currency amounts likely produces bad results (to be tested)

MySQL/MariaDB INNODB

Elasticsearch

Manticore

There is a hard limit of 256 full text fields per index. Additional fields will be slower.

Extra features

Selection guidelines

Tiki 25 and before

  • Small sites, simple functionality: MariaDB/MySQL Full Text Search
  • Medium or large sites, advanced functionality: Elasticsearch

Tiki 26 and up

  • Small sites, simple functionality: MySQL Full Text Search
  • If you are already using Elasticsearch and are happy with it: Elasticsearch
  • Medium or large sites, advanced functionality: Manticore

Speed comparison

Differences of results between engines

index:compare-engines

Legacy

Zend_Search_Lucene (PHP Implementation) was introduced in Tiki7 and later removed Last version: Tiki21.

alias
History Source Comments
Slideshow