Visual Api Training

TRANSFORMATIONS AND ACTIONS
A Visual Guide of the APIhttp://training.databricks.com/visualapi.pdf

Databricks would like to give a special thanks to Jeff Thomspon for contributing 67
visual diagrams depicting the Spark API under the MIT license to the Spark
community.
Jeff’s original, creative work can be found here and you can read more about
Jeff’s project in his blog post.
After talking to Jeff, Databricks commissioned Adam Breindel to further evolve
Jeff’s work into the diagrams you see in this deck.
LinkedIn
Blog: data-frack

making big data simple
Databricks Cloud:
“A unified platform for building Big Data pipelines – from ETL to
Exploration and Dashboards, to Advanced Analytics and Data
Products.”
• Founded in late 2013
• by the creators of Apache Spark
• Original team from UC Berkeley AMPLab
• Raised $47 Million in 2 rounds
• ~55 employees
• We’re hiring!
• Level 2/3 support partnerships with
• Hortonworks
• MapR
• DataStax
(http://databricks.workable.com)

key
RDD Elements
original item
transformed
type
object on driver
RDD
partition(s) A
B
user functions
user input
input
emitted value
Legend

Randomized operation
Legend
Set Theory / Relational operation
Numeric calculation

Operations =
TRANSFORMATIONS
ACTIONS
+

• map
• filter
• flatMap
• mapPartitions
• mapPartitionsWithIndex
• groupBy
• sortBy
= medium
Essential Core & Intermediate Spark OperationsTRANSFORMATIONSACTIONS
General
• sample
• randomSplit
Math / Statistical
= easy
Set Theory / Relational
• union
• intersection
• subtract
• distinct
• cartesian
• zip
• takeOrdered
Data Structure / I/O
• saveAsTextFile
• saveAsSequenceFile
• saveAsObjectFile
• saveAsHadoopDataset
• saveAsHadoopFile
• saveAsNewAPIHadoopDataset
• saveAsNewAPIHadoopFile
• keyBy
• zipWithIndex
• zipWithUniqueID
• zipPartitions
• coalesce
• repartition
• repartitionAndSortWithinPartitions
• pipe
• count
• takeSample
• max
• min
• sum
• histogram
• mean
• variance
• stdev
• sampleVariance
• countApprox
• countApproxDistinct
• reduce
• collect
• aggregate
• fold
• first
• take
• forEach
• top
• treeAggregate
• treeReduce
• forEachPartition
• collectAsMap

= medium
Essential Core & Intermediate PairRDD OperationsTRANSFORMATIONSACTIONS
General
• sampleByKey
Math / Statistical
= easy
Set Theory / Relational Data Structure
• keys
• values
• partitionBy
• countByKey
• countByValue
• countByValueApprox
• countApproxDistinctByKey
• countApproxDistinctByKey
• countByKeyApprox
• sampleByKeyExact
• cogroup (=groupWith)
• join
• subtractByKey
• fullOuterJoin
• leftOuterJoin
• rightOuterJoin
• flatMapValues
• groupByKey
• reduceByKey
• reduceByKeyLocally
• foldByKey
• aggregateByKey
• sortByKey
• combineByKey

vs
narrow wide
each partition of the parent RDD is used by
at most one partition of the child RDD
multiple child RDD partitions may depend
on a single parent RDD partition

“One of the challenges in providing RDDs as an abstraction is choosing a
representation for them that can track lineage across a wide range of
transformations.”
“The most interesting question in designing this interface is how to represent
dependencies between RDDs.”
“We found it both sufficient and useful to classify dependencies into two types:
• narrow dependencies, where each partition of the parent RDD is used by at
most one partition of the child RDD
• wide dependencies, where multiple child partitions may depend on it.”

narrow wide
each partition of the parent RDD is used by
at most one partition of the child RDD
multiple child RDD partitions may depend
on a single parent RDD partition
map, filter union
join w/ inputs
co-partitioned
groupByKey
join w/ inputs not
co-partitioned

TRANSFORMATIONS Core Operations

MAP
User function
applied item by item
RDD: x RDD: y

MAP
RDD: x RDD: y
After map() has been applied…
before after

MAP
RDD: x RDD: y
Return a new RDD by applying a function to each element of this RDD.

MAP
x = sc.parallelize(["b", "a", "c"])
y = x.map(lambda z: (z, 1))
print(x.collect())
print(y.collect())
['b', 'a', 'c']
[('b', 1), ('a', 1), ('c', 1)]
RDD: x RDD: y
x:
y:
map(f, preservesPartitioning=False)
Return a new RDD by applying a function to each element of this RDD
val x = sc.parallelize(Array("b", "a", "c"))
val y = x.map(z => (z,1))
println(x.collect().mkString(", "))
println(y.collect().mkString(", "))

FILTER
Apply user function:
keep item if function
returns true
RDD: x RDD: y
emits
True

FILTER
RDD: x RDD: y
emits
False

FILTER
RDD: x RDD: y
emits
True

FILTER
RDD: x RDD: y
After filter() has been applied…
before after

FILTER
x = sc.parallelize([1,2,3])
y = x.filter(lambda x: x%2 == 1) #keep odd values
print(x.collect())
print(y.collect())
[1, 2, 3]
[1, 3]
RDD: x RDD: y
x:
y:
filter(f)
Return a new RDD containing only the elements that satisfy a predicate
val x = sc.parallelize(Array(1,2,3))
val y = x.filter(n => n%2 == 1)

FLATMAP
RDD: x RDD: y
After flatmap() has been applied…
before after

FLATMAP
RDD: x RDD: y
Return a new RDD by first applying a function to all elements of this RDD, and then flattening the results

FLATMAP
y = x.flatMap(lambda x: (x, x*100, 42))
print(x.collect())
print(y.collect())
[1, 2, 3]
[1, 100, 42, 2, 200, 42, 3, 300, 42]
x:
y:
RDD: x RDD: y
flatMap(f, preservesPartitioning=False)
Return a new RDD by first applying a function to all elements of this RDD, and then flattening the results
val y = x.flatMap(n => Array(n, n*100, 42))

GROUPBY
4 items in RDD
RDD: x
James
Anna
Fred
John

GROUPBY
RDD: x
James
Anna
Fred
John
emits‘J’
J [ “John” ]
RDD: y

F [ “Fred” ]
GROUPBY
RDD: x
James
Anna
Fred
emits‘F’
J [ “John” ]John
RDD: y

[ “Fred” ]
GROUPBY
RDD: x
James
Anna
emits‘A’
J [ “John” ]
A [ “Anna” ]
Fred
John
F
RDD: y

[ “Fred” ]
GROUPBY
RDD: x
James
Anna
emits‘J’
J [ “John”, “James” ]
[ “Anna” ]
Fred
John
F
A
RDD: y

GROUPBY
x = sc.parallelize(['John', 'Fred', 'Anna', 'James'])
y = x.groupBy(lambda w: w[0])
print [(k, list(v)) for (k, v) in y.collect()]
['John', 'Fred', 'Anna', 'James']
[('A',['Anna']),('J',['John','James']),('F',['Fred'])]
RDD: x RDD: y
x:
y:
groupBy(f, numPartitions=None)
Group the data in the original RDD. Create pairs where the key is the output of
a user function, and the value is all items for which the function yields this key.
val x = sc.parallelize(
Array("John", "Fred", "Anna", "James"))
val y = x.groupBy(w => w.charAt(0))

GROUPBYKEY
5 items in RDD
Pair RDD: x
B
B
A
A
A
5
4
3
2
1

GROUPBYKEY
Pair RDD: x
5
4
3
2
1
RDD: y
A [ 2 , 3 , 1 ]
B
B
A
A
A

GROUPBYKEY
Pair RDD: x RDD: y
B [ 5 , 4 ]
A [ 2 , 3 , 1 ]
5
4
3
2
1
B
B
A
A
A

GROUPBYKEY
x = sc.parallelize([('B',5),('B',4),('A',3),('A',2),('A',1)])
y = x.groupByKey()
print(x.collect())
print(list((j[0], list(j[1])) for j in y.collect()))
[('B', 5),('B', 4),('A', 3),('A', 2),('A', 1)]
[('A', [2, 3, 1]),('B',[5, 4])]
RDD: x RDD: y
x:
y:
groupByKey(numPartitions=None)
Group the values for each key in the original RDD. Create a new pair where the
original key corresponds to this collected group of values.
Array(('B',5),('B',4),('A',3),('A',2),('A',1)))
val y = x.groupByKey()

MAPPARTITIONS
RDD: x RDD: y
partitions
A
B
A
B

REDUCEBYKEY VS GROUPBYKEY
val words = Array("one", "two", "two", "three", "three", "three")
val wordPairsRDD = sc.parallelize(words).map(word => (word, 1))
val wordCountsWithReduce = wordPairsRDD
.reduceByKey(_ + _)
.collect()
val wordCountsWithGroup = wordPairsRDD
.groupByKey()
.map(t => (t._1, t._2.sum))
.collect()

REDUCEBYKEY
(a, 1)
(b, 1)
(a, 1)
(b, 1)
(a, 1)
(a, 1) (a, 2)
(b, 2)(b, 1)
(b, 1)
(a, 1)
(a, 1)
(a, 3)
(b, 2)
(a, 1)
(b, 1)
(b, 1)
(a, 1)
(a, 2) (a, 6)
(a, 3)
(b, 1)
(b, 2) (b, 5)
(b, 2)
a b

GROUPBYKEY
(a, 1)
(b, 1)
(a, 1)
(a, 1)
(b, 1)
(b, 1)
(a, 1)
(a, 1)
(a, 1)
(b, 1)
(b, 1)
(a, 1)
(a, 1)
(a, 6)
(a, 1)
(b, 5)
a b
(a, 1)
(a, 1)
(a, 1)
(b, 1)
(b, 1)
(b, 1)
(b, 1)
(b, 1)

MAPPARTITIONS
x:
y:
mapPartitions(f, preservesPartitioning=False)
Return a new RDD by applying a function to each partition of this RDD
A
B
A
B
x = sc.parallelize([1,2,3], 2)
def f(iterator): yield sum(iterator); yield 42
y = x.mapPartitions(f)
# glom() flattens elements on the same partition
print(x.glom().collect())
print(y.glom().collect())
[[1], [2, 3]]
[[1, 42], [5, 42]]

MAPPARTITIONS
x:
y:
mapPartitions(f, preservesPartitioning=False)
Return a new RDD by applying a function to each partition of this RDD
A
B
A
B
Array(Array(1), Array(2, 3))
Array(Array(1, 42), Array(5, 42))
val x = sc.parallelize(Array(1,2,3), 2)
def f(i:Iterator[Int])={ (i.sum,42).productIterator }
val y = x.mapPartitions(f)
// glom() flattens elements on the same partition
val xOut = x.glom().collect()
val yOut = y.glom().collect()

MAPPARTITIONSWITHINDEX
RDD: x RDD: y
partitions
A
B
A
B
input
partition index

x:
y:
mapPartitionsWithIndex(f, preservesPartitioning=False)
Return a new RDD by applying a function to each partition of this RDD,
while tracking the index of the original partition
A
B
A
B
def f(partitionIndex, iterator): yield (partitionIndex, sum(iterator))
y = x.mapPartitionsWithIndex(f)
# glom() flattens elements on the same partition
[[1], [2, 3]]
[[0, 1], [1, 5]]
partition index
B A

x:
y:
mapPartitionsWithIndex(f, preservesPartitioning=False)
Return a new RDD by applying a function to each partition of this RDD,
while tracking the index of the original partition.
A
B
A
B
Array(Array(1), Array(2, 3))
Array(Array(0, 1), Array(1, 5))
partition index
B A
def f(partitionIndex:Int, i:Iterator[Int]) = {
(partitionIndex, i.sum).productIterator
}
val y = x.mapPartitionsWithIndex(f)
// glom() flattens elements on the same partition

SAMPLE
RDD: x RDD: y
1
3
5
4
3
2
1

SAMPLE
x = sc.parallelize([1, 2, 3, 4, 5])
y = x.sample(False, 0.4, 42)
print(x.collect())
print(y.collect())
[1, 2, 3, 4, 5]
[1, 3]
RDD: x RDD: y
x:
y:
sample(withReplacement, fraction, seed=None)
Return a new RDD containing a statistical sample of the original RDD
val x = sc.parallelize(Array(1, 2, 3, 4, 5))
val y = x.sample(false, 0.4)
// omitting seed will yield different output

UNION RDD: x RDD: y
4
3
3
2
1
A
B
C
4
3
3
2
1
A
B
C
RDD: z

UNION
y = sc.parallelize([3,4], 1)
z = x.union(y)
print(z.glom().collect())
[1, 2, 3]
[3, 4]
[[1], [2, 3], [3, 4]]
x:
y:
union(otherRDD)
Return a new RDD containing all items from two original RDDs. Duplicates are not culled.
val y = sc.parallelize(Array(3,4), 1)
val z = x.union(y)
val zOut = z.glom().collect()
z:

5B
JOIN RDD: x RDD: y
4
2B
A
1A
3A

JOIN RDD: x RDD: y
RDD: z
(1, 3)A
5B
4
2B
A
1A
3A

JOIN RDD: x RDD: y
RDD: z
(1, 4)A
(1, 3)A
5B
4
2B
A
1A
3A

JOIN RDD: x RDD: y
(2, 5) RDD: zB
(1, 4)A
(1, 3)A
5B
4
2B
A
1A
3A

JOIN
x = sc.parallelize([("a", 1), ("b", 2)])
y = sc.parallelize([("a", 3), ("a", 4), ("b", 5)])
z = x.join(y)
print(z.collect()) [("a", 1), ("b", 2)]
[("a", 3), ("a", 4), ("b", 5)]
[('a', (1, 3)), ('a', (1, 4)), ('b', (2, 5))]
x:
y:
union(otherRDD, numPartitions=None)
Return a new RDD containing all pairs of elements having the same key in the original RDDs
val x = sc.parallelize(Array(("a", 1), ("b", 2)))
val y = sc.parallelize(Array(("a", 3), ("a", 4), ("b", 5)))
val z = x.join(y)
println(z.collect().mkString(", "))
z:

DISTINCT
RDD: x
4
3
3
2
1
4
3
3
2
1
RDD: y

DISTINCT
RDD: x
4
3
3
2
1
4
3
2
1
RDD: y

DISTINCT
x = sc.parallelize([1,2,3,3,4])
y = x.distinct()
print(y.collect())
[1, 2, 3, 3, 4]
[1, 2, 3, 4]
x:
y:
distinct(numPartitions=None)
Return a new RDD containing distinct items from the original RDD (omitting all duplicates)
val x = sc.parallelize(Array(1,2,3,3,4))
val y = x.distinct()
*
*
*¤
¤

COALESCE
RDD: x
B
C
RDD: y
A
AB

C
COALESCE
RDD: x
B
C
RDD: y
A
AB

COALESCE
x = sc.parallelize([1, 2, 3, 4, 5], 3)
y = x.coalesce(2)
[[1], [2, 3], [4, 5]]
[[1], [2, 3, 4, 5]]
x:
y:
coalesce(numPartitions, shuffle=False)
Return a new RDD which is reduced to a smaller number of partitions
val x = sc.parallelize(Array(1, 2, 3, 4, 5), 3)
val y = x.coalesce(2)
C
B
C
A
AB

KEYBY
RDD: x
James
Anna
Fred
John
RDD: y
J “John”
emits‘J’

KEYBY
RDD: x
James
Anna
‘F’
Fred “Fred”F
RDD: y
J “John”John

James
“Anna”A
KEYBY
RDD: x
Anna
‘A’
Fred
John
“Fred”F
RDD: y
J “John”

J “James”
“Anna”A
KEYBY
RDD: x
James
Anna
emits‘J’
Fred
John
“Fred”F
RDD: y
J “John”

KEYBY
x = sc.parallelize(['John', 'Fred', 'Anna', 'James'])
y = x.keyBy(lambda w: w[0])
print y.collect()
['John', 'Fred', 'Anna', 'James']
[('J','John'),('F','Fred'),('A','Anna'),('J','James')]
RDD: x RDD: y
x:
y:
keyBy(f)
Create a Pair RDD, forming one pair for each item in the original RDD. The
pair’s key is calculated from the value via a user-supplied function.
Array("John", "Fred", "Anna", "James"))
val y = x.keyBy(w => w.charAt(0))

PARTITIONBY
RDD: x
J “John”
“Anna”A
“Fred”F
J “James”

PARTITIONBY
RDD: x
J “John”
“Anna”A
“Fred”F
J “James”
RDD: yRDD: y
J “James”
1

PARTITIONBY
RDD: x
J “John”
“Anna”A
“Fred”F
RDD: yRDD: y
J “James”
“Fred”F
0
J “James”

PARTITIONBY
RDD: x
J “John”
“Anna”A
RDD: yRDD: y
J “James”
“Anna”A
“Fred”F
0
“Fred”F
J “James”

PARTITIONBY
RDD: x
J “John”
RDD: yRDD: y
J “John”
J “James”
“Anna”A
“Fred”F1
“Anna”A
“Fred”F
J “James”

PARTITIONBY
x = sc.parallelize([('J','James'),('F','Fred'),
('A','Anna'),('J','John')], 3)
y = x.partitionBy(2, lambda w: 0 if w[0] < 'H' else 1)
print x.glom().collect()
print y.glom().collect()
[[('J', 'James')], [('F', 'Fred')],
[('A', 'Anna'), ('J', 'John')]]
[[('A', 'Anna'), ('F', 'Fred')],
[('J', 'James'), ('J', 'John')]]
x:
y:
partitionBy(numPartitions, partitioner=portable_hash)
Return a new RDD with the specified number of partitions, placing original
items into the partition returned by a user supplied function

PARTITIONBY
Array(Array((A,Anna), (F,Fred)),
Array((J,John), (J,James)))
Array(Array((F,Fred), (A,Anna)),
Array((J,John), (J,James)))
x:
y:
partitionBy(numPartitions, partitioner=portable_hash)
Return a new RDD with the specified number of partitions, placing original
items into the partition returned by a user supplied function.
import org.apache.spark.Partitioner
val x = sc.parallelize(Array(('J',"James"),('F',"Fred"),
('A',"Anna"),('J',"John")), 3)
val y = x.partitionBy(new Partitioner() {
val numPartitions = 2
def getPartition(k:Any) = {
if (k.asInstanceOf[Char] < 'H') 0 else 1
}
})

ZIP RDD: x RDD: y
3
2
1
A
B
9
4
1
A
B

ZIP RDD: x RDD: y
3
2
1
A
B
4
A
RDD: z
9
4
1
A
B
1 1

ZIP RDD: x RDD: y
3
2
1
A
B
4
A
RDD: z
9
4
1
A
B
2
1
4
1

ZIP RDD: x RDD: y
3
2
1
A
B
4
3
A
B
RDD: z
9
4
1
A
B
2
1
9
4
1

ZIP
x = sc.parallelize([1, 2, 3])
y = x.map(lambda n:n*n)
z = x.zip(y)
print(z.collect()) [1, 2, 3]
[1, 4, 9]
[(1, 1), (2, 4), (3, 9)]
x:
y:
zip(otherRDD)
Return a new RDD containing pairs whose key is the item in the original RDD, and whose
value is that item’s corresponding element (same partition, same index) in a second RDD
val y = x.map(n=>n*n)
val z = x.zip(y)
println(z.collect().mkString(", "))
z:

vs
distributed driver
occurs across the cluster result must fit in driver JVM

GETNUMPARTITIONS
partition(s) A
B
2

x:
y:
getNumPartitions()
Return the number of partitions in RDD
[[1], [2, 3]]
2
GETNUMPARTITIONS
A
B
2
y = x.getNumPartitions()
print(y)
val y = x.partitions.size
println(y)

x:
y:
collect()
Return all items in the RDD to the driver in a single list
[[1], [2, 3]]
[1, 2, 3]
A
B
y = x.collect()
print(y)
val y = x.collect()
println(y)
COLLECT [ ]

4
REDUCE
3
2
1
emits
3
input
6

x:
y:
reduce(f)
Aggregate all the elements of the RDD by applying a user function
pairwise to elements and partial results, and returns a result to the driver
[1, 2, 3, 4]
10
x = sc.parallelize([1,2,3,4])
y = x.reduce(lambda a,b: a+b)
print(x.collect())
print(y)
val x = sc.parallelize(Array(1,2,3,4))
val y = x.reduce((a,b) => a+b)
println(x.collect.mkString(", "))
println(y)
REDUCE
******
*
**
***

AGGREGATE
3
2
1
4
emits
([], 0)
([1], 1)

AGGREGATE
3
2
1
4
([], 0)
([2], 2)
([1], 1)

AGGREGATE
3
2
1
4
([2], 2)
([1], 1)([1,2], 3)

AGGREGATE
3
2
1
4
([2], 2)
([1], 1)([1,2], 3)
([3], 3)
([], 0)

AGGREGATE
3
2
1
4
([2], 2)
([1], 1)([1,2], 3)
([3], 3)
([], 0)
([4], 4)

AGGREGATE
3
2
1
4
([2], 2)
([1], 1)([1,2], 3)
([4], 4)
([3], 3)([3,4], 7)

AGGREGATE
3
2
1
4
([1,2], 3)
([3,4], 7)

AGGREGATE
3
2
1
4
([1,2], 3)
([3,4], 7)
([1,2,3,4], 10)
([1,2,3,4], 10)

aggregate(identity, seqOp, combOp)
Aggregate all the elements of the RDD by:
- applying a user function to combine elements with user-supplied objects,
- then combining those user-defined results via a second user function,
- and finally returning a result to the driver.
seqOp = lambda data, item: (data[0] + [item], data[1] + item)
combOp = lambda d1, d2: (d1[0] + d2[0], d1[1] + d2[1])
x = sc.parallelize([1,2,3,4])
y = x.aggregate(([], 0), seqOp, combOp)
print(y)
AGGREGATE
***
****
**
***
[( ),#]
x:
y:
[1, 2, 3, 4]
([1, 2, 3, 4], 10)

def seqOp = (data:(Array[Int], Int), item:Int) =>
(data._1 :+ item, data._2 + item)
def combOp = (d1:(Array[Int], Int), d2:(Array[Int], Int)) =>
(d1._1.union(d2._1), d1._2 + d2._2)
val x = sc.parallelize(Array(1,2,3,4))
val y = x.aggregate((Array[Int](), 0))(seqOp, combOp)
println(y)
x:
y:
aggregate(identity, seqOp, combOp)
Aggregate all the elements of the RDD by:
- applying a user function to combine elements with user-supplied objects,
- then combining those user-defined results via a second user function,
- and finally returning a result to the driver.
[1, 2, 3, 4]
(Array(3, 1, 2, 4),10)
AGGREGATE
***
****
**
***
[( ),#]

x:
y:
max()
Return the maximum item in the RDD
[2, 4, 1]
4
MAX
4
y = x.max()
print(x.collect())
print(y)
val y = x.max
println(y)
2
1
4
max

x:
y:
sum()
Return the sum of the items in the RDD
[2, 4, 1]
7
SUM
7
y = x.sum()
print(x.collect())
print(y)
val y = x.sum
println(y)
2
1
4
Σ

x:
y:
mean()
Return the mean of the items in the RDD
[2, 4, 1]
2.3333333
MEAN
2.3333333
y = x.mean()
print(x.collect())
print(y)
val y = x.mean
println(y)
2
1
4
x

x:
y:
stdev()
Return the standard deviation of the items in the RDD
[2, 4, 1]
1.2472191
STDEV
1.2472191
y = x.stdev()
print(x.collect())
print(y)
val y = x.stdev
println(y)
2
1
4
σ

COUNTBYKEY
{'A': 1, 'J': 2, 'F': 1}
J “John”
“Anna”A
“Fred”F
J “James”

x:
y:
countByKey()
Return a map of keys and counts of their occurrences in the RDD
[('J', 'James'), ('F','Fred'),
('A','Anna'), ('J','John')]
{'A': 1, 'J': 2, 'F': 1}
COUNTBYKEY
{A: 1, 'J': 2, 'F': 1}
x = sc.parallelize([('J', 'James'), ('F','Fred'),
('A','Anna'), ('J','John')])
y = x.countByKey()
print(y)
val x = sc.parallelize(Array(('J',"James"),('F',"Fred"),
('A',"Anna"),('J',"John")))
val y = x.countByKey()
println(y)

x:
y:
saveAsTextFile(path, compressionCodecClass=None)
Save the RDD to the filesystem indicated in the path
[2, 4, 1]
[u'2', u'4', u'1']
SAVEASTEXTFILE
dbutils.fs.rm("/temp/demo", True)
x.saveAsTextFile("/temp/demo")
y = sc.textFile("/temp/demo")
print(y.collect())
dbutils.fs.rm("/temp/demo", true)
x.saveAsTextFile("/temp/demo")
val y = sc.textFile("/temp/demo")

Visual Api Training

More Related Content

What's hot (20)

Similar to Visual Api Training (20)

More from Spark Summit (20)

Recently uploaded (20)

Visual Api Training