-
Notifications
You must be signed in to change notification settings - Fork 8
/
SparkSummit.scala
140 lines (99 loc) · 2.73 KB
/
SparkSummit.scala
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
package com.github.mrpowers.spark.examples
import org.apache.spark.sql.types._
import org.apache.spark.sql.streaming.Trigger
import org.apache.spark.sql.SaveMode
import org.apache.spark.sql.DataFrame
object SparkSummit extends SparkSessionWrapper {
import spark.implicits._
def createParquetLake() = {
val schema = StructType(
Seq(
StructField("name", StringType, true)
)
)
val sDF = spark.readStream
.schema(schema)
.csv("/my-cool-bucket/csv-lake/data")
sDF
.writeStream
.trigger(Trigger.Once)
.format("parquet")
.option("checkpointLocation", "/my-cool-bucket/parquet-lake/checkpoint")
.start("/my-cool-bucket/parquet-lake/data/incremental")
}
def compactParquetLake() = {
val df = spark
.read
.parquet("/my-cool-bucket/parquet-lake/data/incremental")
df
.coalesce(166)
.write
.mode(SaveMode.Append)
.parquet("/my-cool-bucket/parquet-lake/data/base")
}
def accessDataLake() = {
spark
.read
.parquet("/my-cool-bucket/parquet-lake/data/{incremental,base}")
}
def filteringUnpartitionedLake() = {
val df = spark
.read
.option("header", "true")
.csv("/Users/powers/Documents/tmp/blog_data/people.csv")
df
.where($"country" === "Russia" && $"first_name".startsWith("M"))
.explain()
}
def createPartitionedDataLake() = {
val df = spark
.read
.option("header", "true")
.csv("/Users/powers/Documents/tmp/blog_data/people.csv")
df
.repartition($"country")
.write
.option("header", "true")
.partitionBy("country")
.csv("/Users/powers/Documents/tmp/blog_data/partitioned_lake")
}
def directlyGrabbingPartitions() = {
val russiansDF = spark
.read
.csv("/Users/powers/Documents/tmp/blog_data/partitioned_lake/country=Russia")
russiansDF.where($"first_name".startsWith("M"))
}
def writeEachPartitionAsSingleFile(df: DataFrame) = {
// each partition is single file
df
.repartition($"country")
.write
.option("header", "true")
.partitionBy("country")
.csv("/Users/powers/Documents/tmp/blog_data/partitioned_lake")
}
def writeEachPartitionsAsTonsOfFiles(df: DataFrame) = {
// tons of files get written out
df
.write
.option("header", "true")
.partitionBy("country")
.csv("/Users/powers/Documents/tmp/blog_data/partitioned_lake")
}
def maxHundredFilesPerPartition(df: DataFrame) = {
import org.apache.spark.sql.functions.rand
// max 100 files per partition
df
.repartition(100, $"country", rand)
.write
.option("header", "true")
.partitionBy("country")
.csv("/Users/powers/Documents/tmp/blog_data/partitioned_lake")
}
// create table
spark.sql("CREATE TABLE delta_pond_for_spike USING DELTA LOCATION '/mnt/some-bucket/delta/pond'")
// compaction
spark.sql("OPTIMIZE delta_pond_for_spike")
// clean up files associated with table
spark.sql("VACUUM delta_pond_for_spike")
}