I hope you have Homebrew installed in your mac if not follow this link. I am using a Mac machine, so setup steps related to Mac. I realized its time to meet my future love Spark.
We can do a couple of optimizations but we know those are temporary fixes.
So if we want to share something important to any broad segment users our application goes out of memory because of several reasons like RAM, large object space limit & etc. More than 1 million users visit our company’s Indian cars websites, mobile sites & Apps every day. So, every time based on conditions we need to get the users belongs to those partitions and do UNION if OR condition, INTERSECTION if its AND condition. We are using Cassandra database because our application is write heavy & its time series data. This example segment contains 3 conditions. So, in this case, we need to get segmented users. Let’s say we got a good offer for Maruti Swift car in Hyderabad city we need to send a notification to the users who visited our Web Site or Mobile Site or App last week and did research related to Maruti Swift car and they are from Hyderabad city. So we need to share relevant offer to the user. If we share all offers its literally spamming. Offers differ from city to city, car model to model & it has validity too. Our data team constantly searches for good offers related to Cars. You may ask what will you do with the segmented users or why do you need segmentation? Let me explain with an example related to one of our company’s cool & super website CarDekho. Basically what segmentation does is shortlists the users based on conditions given in segmentation definition like user visited X page and did Y event & etc.
#Download spark and run wordcount example Offline
We recently started offline segmentation support in our current project. I know one day I need to go for a date with Spark but somehow I was postponing for a long time, That day came I am excited about this new journey. I think almost all whoever have a relationship with Big Data will cross Spark path in one way or another way.
0 kommentar(er)
