Skip to content
/ ZHT Public
forked from mierl/ZHT

This is a very light weighted, highly scalable and high performance DHT. It is tuned to meet the needs of HPC community. It is at least 10 to 100 times faster than most of other existing DHT projects. On a real supercomputer test bed, it scales up to 5000 cores without losing much performance.

Notifications You must be signed in to change notification settings

corentone/ZHT

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

81 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

=============================================

=============================================
Software requirement
---------------------------------------------
Google protocol buffers 2.3.0 or newer


=============================================
Compile options

Before compiling, please make sure that Google Protocol Buffer is installed in your system and please edit the variable PROTOBUF_HOME in the makefile with your protocol buffer install location. (a configure step should be added in the future)

Simply compile using: make

Note:If you made any edition to the Protocol Buffer Structure, please recompile it before compiling ZHT by using: make gProto


=============================================
How to use
---------------------------------------------
ZHT is designed to manage file system meta data, if you want to use it storing general purpose data, please make sure it will follow the structure of class Package. Now only plain string is supported inside of Package.

To use ZHT you have to run the ZHT Server Daemon. It is located in the bin directory.

You may want to try some examples and benchmarks, the binaries are located in the examples directory along with the source files.

** If you wish to use ZHT Client in your program you need to import the static library (libzht.a) located in the lib directory. Then, you may need to import header files as well (either all the inc folder or the header files located along with the library in the lib directory).

Establish a ZHT
---------------------------------------------
Note: the $protocol could only be "TCP" or "UDP"

Server:
Simply execute ./server_general $port $memberListFile $configFile $protocol to start a instance of ZHT server node.
How to stop servers: execute ./haltServer $memberListFile $configFile on any node.
Client:
For runing the example, run
./client_general $numOps $memberListFile $configFile $protocol
This will insert, lookup and remove $numOps records in ZHT servers.

Class ZHTClient allows you to create a client object which featured with insert/lookup/remove access to a established ZHT network. 
Before accessing the ZHT, you have to initialize ZHTClient by calling ZHTClient::initialize(string configFilePath, string memberListFilePath).
There is a comprehensive example in client.cpp, please check following functions:
int benchmarkInsert(...)
int benchmarkLookup(...)
int benchmarkRemove(...)



=============================================
Configure file
---------------------------------------------
Configure file must be STRICTLY organized as following format, now it only have two options:

REPLICATION_TYPE=0
NUM_REPLICAS=0

NUM_REPLICAS specify the number of replicas that you want to set, 0 means no replica. For most of applications 3 is adequate.(Now replica has some problem, we’re fixing it, so please set both of them to be 0 for now.)	 



=============================================
Member List File
---------------------------------------------
Member list file should be organized in following way:
hostname1 port1
hostname2 port2
...



=============================================
IMPORTANT NOTES:
1. The class Package must be followed.
2. Make sure the serialized package is smaller than 64KB.




About

This is a very light weighted, highly scalable and high performance DHT. It is tuned to meet the needs of HPC community. It is at least 10 to 100 times faster than most of other existing DHT projects. On a real supercomputer test bed, it scales up to 5000 cores without losing much performance.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • C++ 85.0%
  • C 12.1%
  • Shell 1.9%
  • Python 1.0%