Sesat > Docs + Support > Development platform architecture

Development Platform Architecture

Sesat Software Requirements

See Software Requirements.
See Hardware Requirements.

Server Type Legend

Type Description RAM Disk
Generic Standard 2-CPU pizza-box 2Gb 3x72Gb disk-drives, 10krpm
FAST Standard 2-CPU pizza-box 2Gb Disc-cabinet with 10x72Gb disk-drives, 15krpm
DB Standard 2-CPU pizza-box 4 Gb 3x144Gb disk-drives, 10krpm

Minimal configuration for Proof-of-Concept

The hardware platform must match the following requirements, including "external" services used by SESAT (but not part of SESAT).

Servers Type Description Software Role
1 Generic Search portal server running the search-application part of SESAT. OS: CentOS, RedHat, Debian or other. App-server: Tomcat 5.5. Web-server: Apache2 (not required). SESAT is deployed on Tomcat as one or more .war-files. Required in all SESAT installations.
1 FAST FAST Search Engine for test indices. FAST version 4.1 or 5.0, the latter is preferred. Sufficient for 5-10 mill. documents. Required if not using only external content.
1 DB Database-server for storing and preparing data to be fed to FAST, and also used if installing SESAT ADM DB: MySql 5.0 or 5.1 Not required if data is fed from the filesystem, for example, or if all data is gathered externally.
1 Generic Application Server for SESAT ADM OS: CentOS. App-server: JBoss 4.1 Required if installing SESAT ADM.
1 or 2 Generic/DB 1 server: Integration server for collecting & washing external content. Suggestion: Mule (Open Source) Simple and powerful integration platform. We have no experience with Mule.
    2 servers: Business Objects Data Integrator. FAST 5.0 and BODI Powerful, commercial integration platform, used by

The following illustrates a minimal set of hardware for performing PoC-development:

The blue servers are absolutely mandatory, while the yellow servers are optional depending on your requirements.

Single Site, Permanent Development Platform

The main difference between the "minimal" configuration and this configuration is the introduction of more FAST-servers and load-balanced servers for the search front and SESAT ADM servers. It is important to architecturally match the production platform, hence the load-balanced servers. Also, the number of FAST-servers will grow with:

  • The amount of data to be indexed
  • The number of different indices (for example, it is wise to have one server for Yellow and White pages data, and another server for Wikipedia indexing).

Configuration for multiple sites (,, etc.)

(Use as example)

  • In this example, we will look at the setup used by This includes different components for statistics, documentation, issue-tracking etc. Many of these components are not part of SESAT, but are included for convenience.
 © 2007-2009 Schibsted ASA
Contact us