Snowflake provides a comprehensive data warehousing solution. There's virtually no limit to the number of databases and warehouses you can create (of course, you need snowflake-shaped credits to create and use warehouses). It is a highly scalable solution that complies with all the best data security practices. Snowflake is a well-designed smart product that is undoubtedly the best in its class.
The decoupled storage and computer work seamlessly, seamlessly and simplify scaling. Microclustering was my biggest concern from day one, but frankly it works. You still need some suggestions here or there, but compared to Teradata (or Postgres on a large scale), it's very, very simple and it works. The permissions model is very clean.
The documentation is excellent (especially compared to Teradata or Spark). It also works great as a processing backbone in Spark (see the snowflake controller for the spark catalyst optimizer). Ingestion tools work well for ETL, but probably better for ELT, which means less logic in dumb ETL tools. The query optimizer has hit the mark 95% of the time, and when it isn't, it explains that the plans and tools are excellent (again, we analyze your teradata).
Backup and time travel stuff is very powerful. Due to hot wait times, the ignition time is very fast. And one of the most important things for me is that the extensions of the SQL language are clear and consistent. There is very little magic or incoherent syntax.
But what sets Snowflake apart is its architecture and data exchange capabilities. The Snowflake architecture allows storage and computing to scale independently, so customers can use and pay for storage and computing separately. In addition, the sharing functionality makes it easy for organizations to quickly exchange secure and controlled data in real time. Snowflake is the most popular solution and supports multi-cloud infrastructure environments such as Amazon, Microsoft and GCP.
It is a highly scalable “as a service” cloud data warehouse that allows users to focus on analyzing data instead of spending time managing and adjusting it. It also allows organizations to seamlessly share data with any data consumer, whether a Snowflake customer or not, through reader accounts that can be created directly from the user interface. Snowflake is based on the cloud infrastructure of Amazon Web Services, Microsoft Azure and Google. Compared to traditional data warehouses, Snowflake helped reduce operating costs while maintaining performance.
Snowpipe is Snowflake's continuous data ingestion service, which allows companies to upload data from external storage locations, such as S3, GCP bucket and Azure Blob, as soon as they are available for the stage. To help you do that, here are four keys to establishing a sustainable and adaptable enterprise data warehouse with Snowflake. Snowflake uses secure and highly scalable cloud storage to store structured and semi-structured data such as JSON, AVRO and Parquet. With Snowflake, you can create several separate MPP processing clusters (called Virtual Store) that do not share computing resources with each other and have no impact on performance.
Another question is, where is your data located in Snowflake? Are you on your servers or in your own public cloud? Think about how you expect your data consumers and business applications to leverage Snowflake's data assets. Copies of entire datasets must be made to support each experiment, which can be done with Snowflake's no-copy cloning feature. Industries such as healthcare, financial services, media and entertainment are investing heavily in Snowflake due to its various offerings and benefits mentioned above. Snowflake supports programming languages such as Python, R, Java and C++, which can be used for machine learning (ML).
There is no doubt that reality rejected those beliefs, and Snowflake is almost as safe a bet as one can find if you have to design a new OLAP data storage architecture. .
