Usages

"I want to run a single workload and look at some basic stats for it."

1. Make a command file and save it or pick one from the samples folder.
2. cd into your bssim folder and run ./bssim -wl <path-to-workload>
3. Check your stdout for some basic stats.

"I want to see graphs for the workload I just ran."

1. cd bssim/data
2. Edit config.ini to your liking.
3. python grapher.py
4. By default, graphs are displayed and saved todata/graphs.pdf

"I want to run a workload with a bunch of different latencies and bandwidths and see graphs for it."

1. Edit bssim/data/config.ini to your liking. The latencies and bandwidths will be what the workload is run with and what the graphs will show.
2. Run ./bssim/scripts/latbw.sh <path-to-workload>.

"I want to run my workload with a different latency and bandwidth myself without changing the file or running some script."

1. ./bssim -wl <workload> -lat <latency> -bw <bandwidth>

Commands

Commands work similarly to dhtHell.
Commands are of the syntax node# command arg, where node# can be a single node number or a range in the form [#-#].

Possible commands as of now are:

• put - adds file where arg is the file path
• get - gets file where arg is the file path
• putb - adds block where arg is the contents of the block
• getb - gets block where arg is the contents of the block
• leave - causes nodes to leave network where arg is the number of seconds until the node leaves

There's also a few special commands:

• create_dummy_files <# of files> <file size> - creates a specified number of files in the samples directory with names dummy(n) and then deletes them when the script finishes (see samples/lotsofiles for an example).
• node#->node <latency> <bandwidth> - assigns a latency and bandwidth to the links between the nodes in node# and node (floats, ms and megabits per second).

Workloads

The first line of a workload file should contain comma separated key value pairs.
Example: node_count:20, query_delay:1
The fields you can configure are currently:

• node_count - Number of nodes (integer greater than 1, defaults to 10).
• visibility_delay - Value visibility delay. The time (ms) taken for a value to be visible in the network (integer, defaults to 0).
• query_delay - Routing query delay, the time (ms) taken to receive a response from a routing query (integer, defaults to 0).
• block_size - Block size in bytes (integer, defaults to splitter.DefaultBlockSize).
• deadline - Number of seconds it takes for a GetBlocks request to time out (float, defaults to 60).
• bandwidth - Specifies default bandwidth in megabits/sec for all links. Can be changed using the -> command (float, defaults to 100).
• latency - Specifies default latency in milliseconds for all links. Can be changed using the -> command (float, defaults to 0).
• manual_links - Set this to true if you don't want the nodes all linked at the start of the simulation (defaults to false).

Prometheus

Prometheus metrics are pushed to localhost:8080/metrics. Current metrics collected are:

• file_times_ms
• block_times_ms
• dup_blocks_count