Demystifying ElasticSearch refresh !

By default ElasticSearch is configured to refresh the index every 1 second. This means it will take atleast 1 second to propagate the changes that are made to a document to be made visible during search.But what if we have a requirement to trigger a process only when the search results are made available.

We want index/update or insert request to wait , until the changes made to the documents are available for search before it returns.

refresh parameter is available for these API’s to control when we want our index to refresh and changes made available to user.

  • Setting it to true , refresh = true will cause relevant Primary and Secondary shard , not complete index to be refreshed immediately.
  • Setting it to wait_for, refresh=wait_for will cause the request to wait until the index is refreshed by ElasticSearch based on index.refresh-interval i.e 1 sec by default. Once the index is refreshed the request returns.
  • Setting it to false, refresh=false has no impact on refresh and request returns immediately. It simply means the data will be available in near future.

Note: ElasticSearch will refresh only those shards that have changed, not the entire index

But there is catch in these simple parameters, there are cases that will cause refresh to happen irrespective of the value of refresh parameter you have set

  • if index.max_refresh_listeners which defaults to 1000 is reached. refresh=wait_for will cause the relevant shard to be refreshed immediately.
  • By default GET is realtime i.e each time a GET request is made, it issues index refresh for the appropriate segment. Causing all changes to be made available.

Understanding Co-partitions and Co-Grouping In Spark

The RDD’s in spark are partitioned, using Hash Partitioner by default. Co-partitioned RDD’s uses same partitioner and thus have their data distributed across partitions in same manner.

val data = Array(1, 2, 3, 4, 5)
val rdd1= sc.parallelize(data,10)
val data2 = Array(5,8,9,10,2)

In both of the above defined RDD’s ,same partitioner is used i.e HashPartitioner. HashPartitioner will partition the data in the same way for both RDD’s,same data values in two different RDD will give same Hashvalue. As the number of partitiones specified is also same. These co-partitioned RDD’s reduces the shuffling in network to a great extent. As all the keys required for keyBy transformations will be present in two same partitions of two different RDD’s.

Co-grouping utilizes concept of Co-Partitioning to provide efficient performance improvement when multiple RDD’s are to be joined, over using join again and again. As with every join operation the destination RDD will either have supplied or default value of partitions and the join may or may not require shuffling of two RDD’s that are to be joined based on, if they were co-partitioned and had same number of partitions.


Since rdd1 and rdd2 used same partitioner and also had same number of partitions, the join operation that produces rdd3 will not require any shuffle. But if rdd1 and rdd2 had different number of partitions than the content of rdd with small number of partitions would have been reshuffled.Since number of partitions are not specified, the will depend on default configuration.

Performing another join using rdd3 and rdd4 to create rdd5 will lead to chances of more shuffling. All these shuffling and expensive operations can be avoided by using cogroup when we have multiple RDD’s to be joined.


As the cogroup will create co-partitioned RDD’s

Charts for Data Analysis

Visualizing data efficiently is the first step in understanding the type of distribution( e.g normal distribution) present in available data Set.It also helps in finding skewness,outliers and many other properties present in data , to help us normalize/ clean it before performing any data-analytics on top of it.

Below are the few charts that are most commonly used in Datascience.

It shows the underlying frequency distribution of set of continuous data, divided in intervals bins.The x-axis represents the values present in the data, while the y-axis (and thus the height of each bar) represents the frequency.
Each bin contains the number of occurrences of scores in the data that are contained withing that distribution. The size of bins should be chosen wisely to make sure the resulting graph is able to depict the underlying frequency distribution of data.


Use a histogram when you have numerical data and want to understand the data distribution, including its shape and central tendency

Typically used with large dataset, when we want to find out if there is any relation between variables, provided both are numeric.If there is any relationship between the variables plot across x and y axis the points would scatter across in a way, as if there existed a invisible line.If the relationship is weaker, the dots will be arranged more loosely but still show a tendency for the y variable to either increase or decrease as the x variable increases.If no relationship exists between variables they would be scattered randomly.
“Use this type of graph when you have two numerical variables and are interested in the relationship between them”

Box-and-Whiskers Plot
These are useful when you are comparing numerical data across multiple groups or categories. With a boxplot you can quickly get information about the mean or median of the data, the overall distribution and degree of variation, and the existence of outliers.

“It is especially useful for indicating whether a distribution is skewed and whether there are potential unusual observations (outliers) in the data set. Box and whisker plots are also very useful when large numbers of observations are involved and when two or more data sets are being compared”

Happy reading .. ☺

Understanding Predicates with JAVA8

In mathematics Predicates are functions that can be either True or False. In JAVA8 Predicates are functional interfaces with only functional method test.
As Predicate is defined as a functional interface in JAVA8 it can be used as the assignment target for a lambda expression or method reference.
we can do boolean operations such as and, or, not(negate) with different instances of Predicate. These default methods are –

Default Method Name Explanation
and() It does logical AND of the predicate on which it is called with another predicate. Example: predicate1.and(predicate2)
or() It does logical OR of the predicate on which it is called with another predicate. Example: predicate1.or(predicate2)
negate() It does boolean negation of the predicate on which it is invoked. Example: predicate1.negate()

Following code uses JAVA8 predicate and replaceIf method, now available in collections, to check from the list of transaction to get only those transaction that has a value more than 2lakh and are done online.

Writing Spark Data Frame to HBase

Community behind Spark has made lot of effort’s to make DataFrame Api’s very efficient and scalable. Reading and writing data, to and, from HBase to Spark DataFrame, bridges the gap between complex sql queries that can be performed on spark to that with Key- value store pattern of HBase. The shc connector implements the standard Spark Datasource API, and leverages the Spark Catalyst engine for query optimization.

To map a table in HBase with the table in Spark , we define a Table catalog.Which stores the mapping of keys, column qualifier and column family in HBase with that of table columns in spark.

In order to work more efficiently and making sure not to retrieve unwanted data from region servers.shc connector supports predicate pushdown where the filter conditions are pushed to data as close as possible i.e regionserver in case of HBase.

Support for partition pruning splits the Scan/BulkGet into multiple non-overlapping ranges, only the region servers that has the requested data will perform Scan/BulkGet.

Specifying conditions like Where x >y or Where x

session.sql('Select * from sparkHBase table where x>14567 and x<14568')

will result in a scan operation on HBase with key range between 14567 and 14568

timestamp temp pressure
1501844115 24 760
1501844125 28 800

The above table presented by Spark DataFrame can be saved to HBase by providing the mapping for key, column qualifiers, column name in HBase

def catalog = s"""{
|"table":{"namespace":"default", "name":"ToolLogs"},
|"timestamp":{"cf":"rowkey", "col":"key", "type":"long"},
|"temp":{"cf":"msmt", "col":"temp", "type":"float"},
|"pressure":{"cf":"msmt", "col":"pressure", "type":"float"}

specifying “cf”:”rowkey” for the key column is mandatory though we had msmt as our column family for HBase table,this is how the API is designed to work. Once we have defined the catalog mapping for our Table Catalog, we can store the data in dataframe directly to HBase using

Map(HBaseTableCatalog.tableCatalog -> catalog, HBaseTableCatalog.newTable -> “5”))

Happy Reading …. ☺