CayLey是GO语言写的一个图数据库引擎，支持RESTful API，内置查询编辑器和可视化，支持MQL和JAVASCRIPT查询接口，后端存储支持文件格式，PostgreSQL，mongodb，LevelDB，Bolt。模块化设计，扩展后端存储非常容易。
本文将以PostgreSQL为例，演示一下CayLey的使用。
安装go：
yum install -y go
执行以下命令，克隆cayley和依赖：
mkdir -p ~/cayley && cd ~/cayley
export GOPATH=`pwd`
export PATH=$PATH:~/cayley/bin
mkdir -p bin pkg src/github.com/google
cd src/github.com/google
git clone https://github.com/google/cayley
cd cayley
go get github.com/tools/godep
godep restore
go build ./cmd/cayley
样本数据：
$ ll data
-rw-rw-r--. 1 postgres postgres 26M Jan 17 21:45 30kmoviedata.nq.gz
-rw-rw-r--. 1 postgres postgres 463 Jan 17 21:45 testdata.nq
$ gunzip 30kmoviedata.nq.gz
cayley使用帮助：
$ ./cayley --help
No command --help
Usage:
cayley COMMAND [flags]
Commands:
init Create an empty database.
load Bulk-load a quad file into the database.
http Serve an HTTP endpoint on the given host and port.
dump Bulk-dump the database into a quad file.
repl Drop into a REPL of the given query language.
version Version information.
Flags:
-alsologtostderr=false: log to standard error as well as files
-assets="": Explicit path to the HTTP assets.
-config="": Path to an explicit configuration file.
-db="memstore": Database Backend.
-dbpath="/tmp/testdb": Path to the database.
-dump="dbdump.nq": Quad file to dump the database to (".gz" supported, "-" for stdout).
-dump_type="quad": Quad file format ("json", "quad", "gml", "graphml").
-format="cquad": Quad format to use for loading ("cquad" or "nquad").
-host="127.0.0.1": Host to listen on (defaults to all).
-ignoredup=false: Don't stop loading on duplicated key on add
-ignoremissing=false: Don't stop loading on missing key on delete
-init=false: Initialize the database before using it. Equivalent to running `cayley init` followed by the given command.
-load_size=10000: Size of quadsets to load
-log_backtrace_at=:0: when logging hits line file:N, emit a stack trace
-log_dir="": If non-empty, write log files in this directory
-logstashtype="": enable logstash logging and define the type
-logstashurl="172.17.42.1:5042": logstash url and port
-logtostderr=false: log to standard error instead of files
-port="64210": Port to listen on.
-prof="": Output profiling file.
-quads="": Quad file to load before going to REPL.
-query_lang="gremlin": Use this parser as the query language.
-read_only=false: Disable writing via HTTP.
-replication="single": Replication method.
-stderrthreshold=0: logs at or above this threshold go to stderr
-timeout=30s: Elapsed time until an individual query times out.
-v=0: log level for V logs
-vmodule=: comma-separated list of pattern=N settings for file-filtered logging
假设已有一个PostgreSQL数据库。
IP : 192.168.150.132
PORT : 1921
DBNAME : postgres
USER : digoal
PWD : digoal_pwd
初始化
./cayley init -db=sql -dbpath="postgres://digoal:digoal_pwd@192.168.150.132:1921/postgres?sslmode=disable"
导入数据
./cayley load -quads="data/" -db=sql -dbpath="postgres://digoal:digoal_pwd@192.168.150.132:1921/postgres?sslmode=disable"
50亿测试数据约2TB。
开启repl或http接口服务。
./cayley repl -db=sql -dbpath="postgres://digoal:digoal_pwd@192.168.150.132:1921/postgres?sslmode=disable" -host="0.0.0.0" -port="64210"
或
./cayley http -db=sql -dbpath="postgres://digoal:digoal_pwd@192.168.150.132:1921/postgres?sslmode=disable" -host="0.0.0.0" -port="64210"
使用http接口的图例：
对于PostgreSQL作为后端的场景，优化的手段：
1. 使用GPU加速HASH JOIN和数据扫描。
2. 使用分区表，减少无用块扫描。
3. 其他通用的PG优化手段
如果数据量大到单库的计算资源和IO资源性能支撑不住，可以用Greenplum来实现分布式查询。
查询接口：
Javascript/Gremlin API documentation
图对象
根据节点ID，检索，返回路径
路径对象
路径相交，节点匹配等
查询路径对象
数值转换，等。
