Apache Spark is a multi-language engine for executing data engineering, data science, and machine learning on single-node machines or clusters.
N/A
SingleStore
Score 8.3 out of 10
N/A
SingleStore aims to enable organizations to scale from one to one million customers, handling SQL, JSON, full text and vector workloads in one unified platform.
The in-memory tech is the best in class. If perf. is the hard limit then this is the No.1 to go. When data scales up, the cost of maintaining a big cluster with memory keep increasing (both the infrastructure cost and licensing cost) becomes high. Also the balance between …
I have tried using CSV as a back-end storage, yet I/O is very heavy, direct transit from spark to SingleStore DB (formerly MemSQL) in memory really beats.
Well suited: To most of the local run of datasets and non-prod systems - scalability is not a problem at all. Including data from multiple types of data sources is an added advantage. MLlib is a decently nice built-in library that can be used for most of the ML tasks. Less appropriate: We had to work on a RecSys where the music dataset that we used was around 300+Gb in size. We faced memory-based issues. Few times we also got memory errors. Also the MLlib library does not have support for advanced analytics and deep-learning frameworks support. Understanding the internals of the working of Apache Spark for beginners is highly not possible.
Good for Applications needing instant insights on large, streaming datasets. Applications processing continuous data streams with low latency. When a multi-cloud, high-availability database is required When NOT to Use Small-scale applications with limited budgets Projects that do not require real-time analytics or distributed scaling Teams without experience in distributed databases and HTAP architectures.
It does not release a patch to have back porting; it just releases a new version and stops support; it's difficult to keep up to that pace.
Support engineers lack expertise, but they seem to be improving organically.
Lacks enterprise CDC capability: Change data capture (CDC) is a process that tracks and records changes made to data in a database and then delivers those changes to other systems in real time.
For enterprise-level backup & restore capability, we had to implement our model via Velero snapshot backup.
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
[Until it is] supported on AWS ECS containers, I will reserve a higher rating for SingleStore. Right now it works well on EC2 and serves our current purpose, [but] would look forward to seeing SingleStore respond to our urge of feature in a shorter time period with high quality and security.
SingleStore excels in real-time analytics and low-latency transactions, making it ideal for operational analytics and mixed workloads. Snowflake shines in batch analytics and data warehousing with strong scalability for large datasets. SingleStore offers faster data ingestion and query execution for real-time use cases, while Snowflake is better for complex analytical queries on historical data.
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 deep dives into our most complexed queries and bizarre issues that sometimes only we get comparing to other clients. Our special workload (thousands of Kafka pipelines + high concurrency of queries). The response match to the priority of the request, P1 gets immediate return call. Missing features are treated, they become a client request and being added to the roadmap after internal consideration on all client needs and priority. Bugs are patched quite fast, depends on the impact and feasible temporary workarounds. There is no issue that we haven't got a proper answer, resolution or reasoning
We allowed 2-3 months for a thorough evaluation. We saw pretty quickly that we were likely to pick SingleStore, so we ported some of our stored procedures to SingleStore in order to take a deeper look. Two SingleStore people worked closely with us to ensure that we did not have any blocking problems. It all went remarkably smoothly.
Spark in comparison to similar technologies ends up being a one stop shop. You can achieve so much with this one framework instead of having to stitch and weave multiple technologies from the Hadoop stack, all while getting incredibility performance, minimal boilerplate, and getting the ability to write your application in the language of your choosing.
Greenplum is good in handling very large amount of data. Concurrency in Greenplum was a major problem. Features available in SingleStore like Pipelines and in memory features are not available in Greenplum. Gemfire was not scaling well like SingleStore. Support of both Greenplum and Gemfire was not good. Product team did not help us much like the ones in SingleStore who helped us getting started on our first cluster very fast.
As the overall performance and functionality were expanded, we are able to deliver our data much faster than before, which increases the demand for data.
Metadata is available in the platform by default, like metadata on the pipelines. Also, the information schema has lots of metadata, making it easy to load our assets to the data catalog.
