Welcome to ELT Basics. After watching this video, you will be able to: Describe what an ELT process is, List use cases for ELT processes, and Describe why ELT is an emergent trend. What is an ELT process?​ ​ELT stands for: Extract, ​ Load, and ​ Transform.​ ​ELT is an acronym for a specific automated
data pipeline engineering methodology.​ ELT is similar to ETL in that similar stages are involved but the order in which
they are performed is different. For ELT processes, data is acquired and directly
loaded, as-is, into its destination environment. From its new home, usually a sophisticated
analytics platform such as a data lake, it can be transformed on demand
and however users wish. ​ Like ETL, the first stage in
the ELT process is Extraction. The Extraction process obtains the data from all sources and reads the data, often in an
asynchronous fashion, into an application.​ The Loading process takes the raw data as-is,and loads it into its new environment, where modern
analytics tools can then be used directly. The Transformation process for ELT is much
more dynamic than it is for conventional ETL. Modern analytics tools in
the destination environment enable interactive, on-demand exploration and visualization of your data, including advanced
analytics such as modelling and prediction. Use cases for ELT processes
typically fall within the high-performance computing and Big Data realms. Cases include: Dealing with the massive swings in scale that
come with implementing Big Data products, Calculating real time analytics
on streaming Big Data, and Bringing together data sources that are
highly distributed around the globe. In terms of speed, moving data is usually
more of a bottleneck than processing it, so the less you move it, the better. Therefore, ELT may be your best bet when you want flexibility in building a suite of
data products from the same sources. Why is ELT emerging? Firstly, cloud computing solutions are evolving at
tremendous rates due to the demands of Big Data. They can easily handle huge
amounts of asynchronous data which can be highly distributed around the world. Cloud computing resources are practically
unlimited, and they can scale on demand. Unlike traditional on-premises hardware, you
only pay for the computing resources you use. You don’t have to worry about underutilizing
resources, that is, overspending on equipment. With ELT, you have a clean separation
between moving data and processing data. Of course, cloud computing is equally prepared to handle the most challenging cases
for either of these two tasks. There may be many reasons to transform
your data and just as many ways to do it. Thus, ELT is a flexible option that enables a variety of applications
from the same source of data. Because you are working with a replica of the
source data, there is no information loss. Many kinds of transformations can lead to
information loss, and if these happen somewhere upstream in the pipeline, it may be a long
time before you can have a change request met. Worse yet, the information may be forever
lost if the raw data is not stored. In this video, you learned that: ELT processes are used for cases where
flexibility, speed, and scalability are important. Cloud-based analytics platforms
are ideally suited for handling Big Data and ELT processes
in a cost-efficient manner. ELT is an emerging trend mainly because
cloud platform technologies are enabling it.