Hailstorm

Scala implementation of Hailstorm: Disaggregated Compute and Storage for Distributed LSM-based Databases.

What is it?

If you have to ask, you probably should read the paper first.

Note: this is the filesystem part of Hailstorm. You will also need the modified RocksDB to run the experiments in the paper.

Requirements

• Scala
• SBT
• Up-to-date JVM

Getting started

Clone the project locally:

$ git clone https://github.com/bindscha/hailstorm.git

Build:

$ sbt compile

Compilation can take a while as SBT fetches all necessary dependencies.

Run:

$ sbt run

To specify CLI options:

$ sbt
sbt:hailstorm> run <options>

Available CLI options:

-b, --blacklist  <arg>...      Blacklist matching files and folders
-p, --blacklist-path  <arg>    Where to put blacklisted files (should be outside HailstormFS mountpoint)
-i, --clear-on-init            Drop persistent file mapping at startup
-c, --client                   Run client
-d, --daemon                   Run daemon
-f, --file-mapping-db  <arg>   Where to put the persistent file mapping database
-o, --fuse-opts  <arg>...      FUSE options
--me  <arg>                    Node ID
-m, --mountpoint  <arg>        Where to mount the filesystem
-l, --offloading               Enable Hailstorm Agent for compaction  offloading
-v, --verbose                  Enable debug logging
-h, --help                     Show help message
--version                      Show version of this program

License

Please see the file called LICENSE.md.

Contact

• Laurent Bindschaedler [email protected]