Apache Spark

DataFrame as a distributed collection of data: easy for developers to implement algorithms and formulas.

Calculation in-memory.

Cluster to distribute large data of calculation.

It would be great if Apache Spark could provide a native database to manage all file info of saved parquet.

Machine learning on big data

Stream processing

Lakehouse with Delta

Indexing

Mllib

Streaming

Apache Spark makes processing very large data sets possible. It handles these data sets in a fairly quick manner.

Apache Spark does a fairly good job implementing machine learning models for larger data sets.

Apache Spark seems to be a rapidly advancing software, with the new features making the software ever more straight-forward to use.

Apache Spark requires some advanced ability to understand and structure the modeling of big data. The software is not user-friendly.

The graphics produced by Apache Spark are by no means world-class. They sometimes appear high-schoolish.

Apache Spark takes an enormous amount of time to crunch through multiple nodes across very large data sets. Apache Spark could improve this by offering the software in a more interactive programming environment.

quick

utilized CPU cores

trendy

lack of support

memory hungry

slow on wide data

Spark is very fast compered to other frameworks because it works in cluster mode and use distributed processing and computation frameworks internally

Robust and fault tolerant

Open source

Can source data from multiple data sources

No Dataset API support in python version of spark

Apache Spark job run UI can have more meaningful information

Spark errors can provide more meaningful information when a job is failed

Great computing engine for solving complex transformative logic

Useful for understanding data and doing data analytical work

Gives us a great set of libraries and api to solve day-to-day problems

High learning curve

Complexity

More documentation

More developer support

More educational videos

Speed: Apache Spark has great performance for both streaming and batch data

Easy to use: the object oriented operators make it easy and intuitive.

Multiple language support

Fault tolerance

Cluster managment

Supports DF, DS, and RDDs

Hard to learn, documentation could be more in-depth.

Due to it's in-memory processing, it can take a large consumption of memory.

Poor data visualization, too basic.

DataFrames, DataSets, and RDDs.

Spark has in-built Machine Learning library which scales and integrates with existing tools.

The data processing done by Spark comes at a price of memory blockages, as in-memory capabilities of processing can lead to large consumption of memory.

The caching algorithm is not in-built in Spark. We need to manually set up the caching mechanism.

Cluster management for ETL.

Data processing engine for our data lake.

You still need Hive or other HDFS to store information.

Security is behind compared to MapReduce.

Very good tool to process big datasets.

Inbuilt fault tolerance.

Supports multiple languages.

Supports advanced analytics.

A large number of libraries available -- GraphX, Spark SQL, Spark Streaming, etc.

Very slow with smaller amounts of data.

Expensive, as it stores data in memory.

Customizable, it integrates with Jupyter notebooks which was really helpful for our team.

Easy to use and implement.

It allows us to quickly build microservices.

Release cycles can be faster.

Sometimes it kicked some of the users out due to inactivity.

in memory data engine and hence faster processing

does well to lay on top of hadoop file system for big data analytics

very good tool for streaming data

could do a better job for analytics dashboards to provide insights on a data stream and hence not have to rely on data visualization tools along with spark

also there is room for improvement in the area of data discovery

It performs a conventional disk-based process when the data sets are too large to fit into memory, which is very useful because, regardless of the size of the data, it is always possible to store them.

It has great speed and ability to join multiple types of databases and run different types of analysis applications. This functionality is super useful as it reduces work times

Apache Spark uses the data storage model of Hadoop and can be integrated with other big data frameworks such as HBase, MongoDB, and Cassandra. This is very useful because it is compatible with multiple frameworks that the company has, and thus allows us to unify all the processes.

Increase the information and trainings that come with the application, especially for debugging since the process is difficult to understand.

It should be more attentive to users and make tutorials, to reduce the learning curve.

There should be more grouping algorithms.

Rich APIs for data transformation making for very each to transform and prepare data in a distributed environment without worrying about memory issues

Faster in execution times compare to Hadoop and PIG Latin

Easy SQL interface to the same data set for people who are comfortable to explore data in a declarative manner

Interoperability between SQL and Scala / Python style of munging data

Documentation could be better as I usually end up going to other sites / blogs to understand the concepts better

More APIs are to be ported to MLlib as only very few algorithms are available at least in clustering segment

Easy ELT Process

Easy clustering on cloud

Amazing speed

Batch & real time processing

Debugging is difficult as it is new for most people

There are fewer learning resources

Machine Learning.

Data Analysis

WorkFlow process (faster than MapReduce).

SQL connector to multiple data sources

Memory management. Very weak on that.

PySpark not as robust as scala with spark.

spark master HA is needed. Not as HA as it should be.

Locality should not be a necessity, but does help improvement. But would prefer no locality

We used to make our batch processing faster. Spark is faster in batch processing than MapReduce with it in memory computing

Spark will run along with other tools in the Hadoop ecosystem including Hive and Pig

Spark supports both batch and real-time processing

Apache Spark has Machine Learning Algorithms support

Consumes more memory

Difficult to address issues around memory utilization

Expensive - In-memory processing is expensive when we look for a cost-efficient processing of big data

Ease of use, the Spark API allows for minimal boilerplate and can be written in a variety of languages including Python, Scala, and Java.

Performance, for most applications we have found that jobs are more performant running via Spark than other distributed processing technologies like Map-Reduce, Hive, and Pig.

Flexibility, the frameworks comes with support for streaming, batch processing, sql queries, machine learning, etc. It can be used in a variety of applications without needing to integrate a lot of other distributed processing technologies.

Resource heavy, jobs, in general, can be very memory intensive and you will want the nodes in your cluster to reflect that.

Debugging, it has gotten better with every release but sometimes it can be difficult to debug an error due to ambiguous or misleading exceptions and stack traces.

Spark uses Scala which is a functional programming language and easy to use language. Syntax is simpler and human readable.

It can be used to run transformations on huge data on different cluster parallelly. It automatically optimizes the process to get output efficiently in less time.

It also provides machine learning API for data science applications and also Spark SQL to query fast for data analysis.

I also use Zeppelin online tool which is used to fast query and very helpful for BI guys to visualize query outputs.

Data visualization.

Waiting for Web Development for small apps to be started with Spark as backbone middleware and HDFS as data retrieval file system.

Transformations and actions available are limited so must modify API to work for more features.

Great APIs and tools.

Scale.

Speed for iterative algorithms.

No true streaming.

Lack of strongly typed yet convenient APIs.

It makes the ETL process very simple when compared to SQL SERVER and MYSQL ETL tools.

It's very fast and has many machine learning algorithms which can be used for data science problems.

It is easily implemented on a cloud cluster.

The initialization and spark context procedures.

Running applications on a cluster is not well documented anywhere, some applications are hard to debug.

Debugging and Testing are sometimes time-consuming.

Scale from local machine to full cluster. You can run a standalone, single cluster simply by starting up a Spark Shell or submitting an application to test an algorithm, then it quickly can be transferred and configured to run in a distributed environment.

Provides multiple APIs. Most people I know use Python and/or Java as their main programming language. Data scientists who are familiar with NumPy and SciPy can quickly become comfortable with Spark, while Java developers would best served using Java 8 and the new features that it provides. Scala, on the other hand, is a mix between the Java and Python styles of writing Spark code, in my opinion.

Plentiful learning resources. The Learning Spark book is a good introduction to the mechanics of Spark although written for Spark 1.3, and the current version is 2.0. The GitHub repository for the book contains all the code examples that are discussed, plus the Spark website is also filled with useful information that is simple to navigate.

For data that isn't truly that large, Spark may be overkill when the problem could likely be solved on a computer with reasonable hardware resources. There doesn't seem to be a lot of examples for how a Spark task would otherwise be implemented in a different library; for instance scikit-learn and NumPy rather than Spark MLlib.