Apache Spark vs. Amazon Redshift

Apache Spark has rich APIs for regular data transformations or for ML workloads or for graph workloads, whereas other systems may not such a wide range of support. Choose it when you need to perform data transformations for big data as offline jobs, whereas use MongoDB-like distributed database systems for more realtime queries.

Incentivized

If the number of connections is expected to be low, but the amounts of data are large or projected to grow it is a good solutions especially if there is previous exposure to PostgreSQL. Speaking of Postgres, Redshift is based on several versions old releases of PostgreSQL so the developers would not be able to take advantage of some of the newer SQL language features. The queries need some fine-tuning still, indexing is not provided, but playing with sorting keys becomes necessary. Lastly, there is no notion of the Primary Key in Redshift so the business must be prepared to explain why duplication occurred (must be vigilant for)

Incentivized

• 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.

Incentivized

• Redshift is fully managed. Small teams do not have the resources to maintain a cluster. CloudWatch metrics are provided out-of-the-box, and it is easy to configure alarms.
• Redshift's console allows you to easily inspect and manage queries, and manage the performance of the cluster.
• Redshift is ubiquitous; many products (e.g., ETL services) integrate with it out-of-the-box.
• Writing .csvs to S3 and querying them through Redshift Spectrum is convenient.

Incentivized

• 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

Incentivized

• It could benefit from adding data integrity and programming tools common to other database management systems.
• Amazon Redshift is based on PostgreSQL 8.0.2. That version of PostgreSQL was released in December 2006. While PostgreSQL was much improved since then, the new features were not implemented in Redshift. Many basic features are missing from it.
• Primary keys can be declared but not enforced. Referential integrity (foreign keys) can be declared but not enforced. UNIQUE and CHECK constraints are not supported and cannot be declared.
• IDENTITY can be declared on a column, and Redshift will put unique values into it. However: IDENTITY values in the newly inserted rows won’t be incremental or sequential. To implement a sequential number, you need to write your own custom code.
• There are no stored procedures in Redshift. We are writing SQL script files, and then parsing and running them one statement at a time from a Python program. This also enabled us to implement execution-time error logging.
• In SQL scripts, to check for the row count of affected rows, a complicated join query against some system tables or views has to be executed.
• Data Control Language (DCL) does not exist. No statements like IF, WHILE, DO, RAISERROR, etc.
• On performance of views… Views do not “pass-through” a query parameter which is a potential problem for performance.
• When selecting against a view with the WHERE clause outside of the view, the inner query of the view will be executed first without consideration for the WHERE clause, and only then the WHERE clause will be applied.
• Certain clauses of SQL work many times faster than other clauses. So be careful and test your statements for performance earlier rather than later, especially if working with a large data set.
• There was a situation when DELETE FROM JOIN was unacceptably slow. Replacing JOIN with the USING clause made DELETE instantaneous.

Incentivized

Capacity of computing data in cluster and fast speed.

If the team looking to use Apache Spark is not used to debug and tweak settings for jobs to ensure maximum optimizations, it can be frustrating. However, the documentation and the support of the community on the internet can help resolve most issues. Moreover, it is highly configurable and it integrates with different tools (eg: it can be used by dbt core), which increase the scenarios where it can be used

Incentivized

Overall it serves all our aspects of data management like data cleaning, data manipulation, and data reporting on the cloud platform. We can create stored procedures and triggers in it very easily as all the options are self suggested in it. We can easily attach the results of ARS to the other tools as well for drawing the statistical results.

Incentivized

1. It integrates very well with scala or python. 2. It's very easy to understand SQL interoperability. 3. Apache is way faster than the other competitive technologies. 4. The support from the Apache community is very huge for Spark. 5. Execution times are faster as compared to others. 6. There are a large number of forums available for Apache Spark. 7. The code availability for Apache Spark is simpler and easy to gain access to. 8. Many organizations use Apache Spark, so many solutions are available for existing applications.

The support was great and helped us in a timely fashion. We did use a lot of online forums as well, but the official documentation was an ongoing one, and it did take more time for us to look through it. We would have probably chosen a competitor product had it not been for the great support

Incentivized

We used Surprise Kit for one of the other research works. It is more fine-tuned to Recommendation systems and their algorithms. Apache Spark has MLlib for majority of ML problems. Where as software like Surprse Kit - it suitable for a specific task of Recommendations only

Incentivized

We evaluated [Amazon] Redshift vs BigQuery vs Amazon EMR, back in 2014. Back then BigQuery cost was slightly higher than that of [Amazon] Redshift price structure. Amazon EMR, needs lots more management (Admin tasks) and EMR is designed to be ephemeral and not designed to be a data store. [Amazon] Redshift was ideal with the price structure, performance and ROI[.]

Incentivized

• Faster turn around on feature development, we have seen a noticeable improvement in our agile development since using Spark.
• Easy adoption, having multiple departments use the same underlying technology even if the use cases are very different allows for more commonality amongst applications which definitely makes the operations team happy.
• Performance, we have been able to make some applications run over 20x faster since switching to Spark. This has saved us time, headaches, and operating costs.

Incentivized

• It allows for an almost seamless integration of our data which can then be used by other departments for analytical purposes.
• No in house resources are needed for keeping the data alive and performing backup/migration tasks of the data in its end state.

Incentivized