SLG is an efficient distributed load injector for Web server. SLG is designed using a master/slave scheme, i.e. a master node synchronizes a set of load injection nodes (each simulating multiple HTTP clients) and collects their results. SLG is a closed-loop load injector (open-loop mode is experimental).
SLG supports several file access patterns: single file, SPECweb99, SPECweb2005 Support, SPECweb09 Banking Workload.
SLG also has a very experimental support of the memcached protocol, as well as FastCGI protocol (tested for PHP-FPM).
Howto to run:
- Select the file access pattern is master_slave.h
- Compile
- Launch slg on each client node using ./slg
- Create a configuration file for the master (see conf/ directory for examples)
- Launch the master using ./slg_master -f
- Use libgsl for statistics