Hadoop vs Spark performance comparison

1. Kmeans

Data: three-dimensional data produced by oneself, each around 8 vertices of a square

{0, 0, 0}, {0, 10, 0}, {0, 0, 10}, {0, 10, 10},

{10, 0, 0}, {10, 0, 10}, {10, 10, 0}, {10, 10, 10}

Point number

189,918,082 (190 million three-dimensional points)

Capacity

10GB

HDFS Location

/user/lijiexu/kmeans/square-10gb.txt

Program logic:

Read block to memory on HDFs, each block converted to Rdd, which contains vectors.

Then the map operation is performed on the RDD, the class number of each vector (point) is extracted, and the output (K,V) is (class, (point,1)) to form a new rdd.

Then, before you reduce it, each new RDD is combine and the center of each class is calculated within the RDD. So that the output of each RDD is only a maximum of k kv pairs.

Finally, reduce gets the new Rdd (the key to the content is the center of the Class,value and the final center after the map.

Upload first to HDFs, then run on master

root@master:/opt/spark#/run spark.examples.http://www.aliyun.com/zixun/aggregation/13383.html ">SparkKMeans master@master:5050 Hdfs://master:9000/user/lijiexu/kmeans/square-10gb.txt 8 2.0

Iterative execution of the Kmeans algorithm.

Altogether 160 tasks. (160 * 64MB = 10GB)

Leverages 32 CPU CORES,18.9GB memory.

Memory consumption for each machine is 4.5GB (total 40GB) (Points data 10gb*2,map (K, V) => (int, (vector, 1)) (approximately 10GB)

Final results:

0.505246194 s

Final Centers:map (5-> (13.997101228817169, 9.208875044622895, -2.494072457488311), 8-> (-2.33522333047955, 9.128892414676326, 1.7923150585737604), 7-> (8.658031587043952, 2.162306996983008, 17.670646829079146), 3-> ( 11.530.54433698268, 0.17834347219956842, 9.224352885937776), 4-> (12.722903153986868, 8.812883284216143, 0.6564509961064319), 1-> (6.458644369071984, 11.345681702383024, 7.041924994173552), 6-> (12.887793408866614,- 1.5189406469928937, 9.526393664105957), 2-> (2.3345459304412164, 2.0173098597285533, 1.4772489989976143))

50mb/s 10GB => 3.5min

10mb/s 10GB => 15min

on 20GB data

Point number

377,370,313 (370 million three-dimensional points)

Capacity

20GB

HDFS Location

/user/lijiexu/kmeans/square-20gb.txt

To run the test command:

root@master:/opt/spark#/run Spark.examples.SparkKMeans master@master:5050 Hdfs://master:9000/user/lijiexu/kmeans /square-20gb.txt 8 2.0 | Tee Mylogs/sqaure-20gb-kmeans.log

Get clustering Result:

Final Centers:map (5-> ( -0.47785701742763115, -1.5901830956323306, -0.18453046159033773), 8-> ( 1.1073911553593858, 9.051671594514225, -0.44722211311446924), 7-> (1.4960397239284795, 10.173412443492643,- 1.7932911100570954), 3-> ( -1.4771114031182642, 9.046878176063172, -2.4747981387714444), 4-> (- 0.2796747780312184, 0.06910629855122015, 10.268115903887612), 1-> (10.467618592186486,-1.168580362309453,- 1.0462842137817263), 6-> (0.7569895433952736, 0.8615441990490469, 9.552726007309518), 2-> (10.807948500515304, -0.5368803187391366, 0.04258123037074164)

Basically, 8 centers.

Memory consumption: (approximately 5.8GB per node), a total of about 50GB.

Memory Analysis:

20GB raw data, 20GB map output

Number of iterations

Time

1

108 S

2

0.93 s

12/06/05 11:11:08 INFO Spark. Cachetracker:looking for RDD partition 2:302

12/06/05 11:11:08 INFO Spark. Cachetracker:found Partition in cache!