Deploying the Aurelius Graph Cluster

jieforest

Once in the machine via ssh, Titan 0.1.0 is downloaded, unzipped, and the Gremlin console is started.

1. 01.ubuntu@ip-10-117-55-34:~$ ssh 54.242.14.83
   
2. 02....ubuntu@ip-10-12-27-208:~$ wget <a href="https://github.com/downloads/thinkaurelius/titan/titan-0.1.0.zip">https://github.com/downloads/thinkaurelius/titan/titan-0.1.0.zip</a>
   
3. 03.ubuntu@ip-10-12-27-208:~$ sudo apt-get install unzip
   
4. 04.ubuntu@ip-10-12-27-208:~$ unzip titan-0.1.0.zip
   
5. 05.ubuntu@ip-10-12-27-208:~$ cd titan-0.1.0/
   
6. 06.ubuntu@ip-10-12-27-208:~/titan-0.1.0$ bin/gremlin.sh
   
7. 07.
   
8. 08.\,,,/
   
9. 09.(o o)
   
10. 10.-----oOOo-(_)-oOOo-----
    
11. 11.gremlin>

复制代码

jieforest

A toy 1 million vertex/edge graph is loaded into Titan using the Gremlin/Groovy script below (simply cut-and-paste the source into the Gremlin console and wait approximately 3 minutes).

The code implements a preferential attachment algorithm. For an explanation of this algorithm, please see the second column of page 33 in Mark Newman‘s article The Structure and Function of Complex Networks.

1. 01.// connect Titan to HBase in batch loading mode
   
2. 02.conf = new BaseConfiguration()
   
3. 03.conf.setProperty('storage.backend','hbase')
   
4. 04.conf.setProperty('storage.hostname','localhost')
   
5. 05.conf.setProperty('storage.batch-loading','true');
   
6. 06.g = TitanFactory.open(conf)
   
7. 07.
   
8. 08.// preferentially attach a growing vertex set
   
9. 09.size = 1000000; ids = [g.addVertex().id]; rand = new Random();
   
10. 10.(1..size).each{
    
11. 11.v = g.addVertex();
    
12. 12.u = g.v(ids.get(rand.nextInt(ids.size())))
    
13. 13.g.addEdge(v,u,'linked');
    
14. 14.ids.add(u.id);
    
15. 15.ids.add(v.id);
    
16. 16.if(it % 10000 == 0) {
    
17. 17.g.stopTransaction(SUCCESS)
    
18. 18.println it
    
19. 19.}
    
20. 20.}; g.shutdown()

复制代码

jieforest

Batch Analytics with Faunus



Faunus is a Hadoop-based graph computing framework. It supports performant global graph analyses by making use of sequential reads from disk (see The Pathologies of Big Data).

Faunus provides connectivity to Titan/HBase, Titan/Cassandra, any Rexster-fronted graph database, and to text/binary files stored in HDFS.

From the 1 zookeeper+hadoop-namenode+hadoop-jobtracker+hbase-master machine, Faunus 0.1-alpha is downloaded and unzipped.

jieforest

The provided titan-hbase.properties file should be updated withhbase.zookeeper.quorum=10.12.27.208 instead of localhost. The IP address 10.12.27.208 is provided by ~/.whirr/agc/instances on agc-master. Finally, the Gremlin console is started.

1. 01.ubuntu@ip-10-12-27-208:~$ wget <a href="https://github.com/downloads/thinkaurelius/faunus/faunus-0.1-alpha.zip">https://github.com/downloads/thinkaurelius/faunus/faunus-0.1-alpha.zip</a>
   
2. 02.ubuntu@ip-10-12-27-208:~$ unzip faunus-0.1-alpha.zip
   
3. 03.ubuntu@ip-10-12-27-208:~$ cd faunus-0.1-alpha/
   
4. 04.ubuntu@ip-10-12-27-208:~/faunus-0.1-alpha$ vi bin/titan-hbase.properties
   
5. 05.ubuntu@ip-10-12-27-208:~/faunus-0.1-alpha$ bin/gremlin.sh
   
6. 06.
   
7. 07.\,,,/
   
8. 08.(o o)
   
9. 09.-----oOOo-(_)-oOOo-----
   
10. 10.gremlin>

复制代码

jieforest

A few example Faunus jobs are provided below. The final job on line 9 generates an in-degree distribution. The in-degree of a vertex is defined as the number of incoming edges to the vertex.

The outputted result states how many vertices (second column) have a particular in-degree (first column). For example, 167,050 vertices have only 1 incoming edge.

1. The provided titan-hbase.properties file should be updated withhbase.zookeeper.quorum=10.12.27.208 instead of localhost. The IP address 10.12.27.208 is provided by ~/.whirr/agc/instances on agc-master. Finally, the Gremlin console is started.
   
2. 01.ubuntu@ip-10-12-27-208:~$ wget <a href="https://github.com/downloads/thinkaurelius/faunus/faunus-0.1-alpha.zip">https://github.com/downloads/thinkaurelius/faunus/faunus-0.1-alpha.zip</a>
   
3. 02.ubuntu@ip-10-12-27-208:~$ unzip faunus-0.1-alpha.zip
   
4. 03.ubuntu@ip-10-12-27-208:~$ cd faunus-0.1-alpha/
   
5. 04.ubuntu@ip-10-12-27-208:~/faunus-0.1-alpha$ vi bin/titan-hbase.properties
   
6. 05.ubuntu@ip-10-12-27-208:~/faunus-0.1-alpha$ bin/gremlin.sh
   
7. 06.
   
8. 07.\,,,/
   
9. 08.(o o)
   
10. 09.-----oOOo-(_)-oOOo-----
    
11. 10.gremlin>
    
12. A few example Faunus jobs are provided below. The final job on line 9 generates an in-degree distribution. The in-degree of a vertex is defined as the number of incoming edges to the vertex. The outputted result states how many vertices (second column) have a particular in-degree (first column). For example, 167,050 vertices have only 1 incoming edge.
    
13. 01.gremlin> g = FaunusFactory.open('bin/titan-hbase.properties')
    
14. 02.==>faunusgraph[titanhbaseinputformat]
    
15. 03.gremlin> g.V.count() // how many vertices in the graph?
    
16. 04.==>1000001
    
17. 05.gremlin> g.E.count() // how many edges in the graph?
    
18. 06.==>1000000
    
19. 07.gremlin> g.V.out.out.out.count() // how many length 3 paths are in the graph?
    
20. 08.==>988780
    
21. 09.gremlin> g.V.sideEffect('{it.degree = it.inE.count()}').degree.groupCount// what is the graph's in-degree distribution?
    
22. 10.==>1 167050
    
23. 11.==>10    2305
    
24. 12.==>100   6
    
25. 13.==>108   3
    
26. 14.==>119   3
    
27. 15.==>122   3
    
28. 16.==>133   1
    
29. 17.==>144   2
    
30. 18.==>155   1
    
31. 19.==>166   2
    
32. 20.==>18    471
    
33. 21.==>188   1
    
34. 22.==>21    306
    
35. 23.==>232   1
    
36. 24.==>254   1
    
37. 25.==>...
    
38. 26.gremlin>

复制代码

jieforest

To conclude, the in-degree distribution result is pulled from Hadoop’s HDFS (stored in output/job-0). Next, scp is used to download the file to agc-master and then again to download the file to a local machine (e.g. a laptop). If the local machine has R installed, then the file can be plotted and visualized (see the final diagram below).

The log-log plot demonstrates the known result that the preferential attachment algorithm generates a graph with a power-law degree distribution (i.e. “natural statistics”).

1. 01.ubuntu@ip-10-12-27-208:~$ hadoop fs -getmerge output/job-0 distribution.txt
   
2. 02.ubuntu@ip-10-12-27-208:~$ head -n5 distribution.txt
   
3. 03.1   167050
   
4. 04.10  2305
   
5. 05.100 6
   
6. 06.108 3
   
7. 07.119 3
   
8. 08.ubuntu@ip-10-12-27-208:~$ exit
   
9. 09....
   
10. 10.ubuntu@ip-10-117-55-34:~$ scp 54.242.14.83:~/distribution.txt .
    
11. 11.ubuntu@ip-10-117-55-34:~$ exit
    
12. 12....
    
13. 13.~$ scp ubuntu@ec2-184-72-209-80.compute-1.amazonaws.com:~/distribution.txt .
    
14. 14.~$ r
    
15. 15.> t = read.table('distribution.txt')
    
16. 16.> plot(t,log='xy',xlab='in-degree',ylab='frequency')

复制代码

jieforest

Conclusion

The Aurelius Graph Cluster is used for processing massive-scale graphs, where massive-scaledenotes a graph so large it does not fit within the resource confines of a single machine. In other words, the Aurelius Graph Cluster is all about Big Graph Data. The two cluster technologies explored in this post were Titan and Faunus.

They serve two distinct graph computing needs. Titan supports thousands of concurrent real-time, topologically local graph interactions. Faunus, on the other hand, supports long running, topologically global graph analyses. In other words, they provide OLTP andOLAP functionality, respectively.

jieforest

over.