Snowflake Expert Services
We have listed some common challenges that our customers experience. We can help you as well.
We get you up and running on Snowflake cloud data warehouse
We tailor our standard process for cloud data warehouse implementations to your individual situation
- Define and document process for gathering requirements.
- Define and document process for data analysis.
- Define and document governance process, e.g. naming conventions, metadata management, data modelling (and tools), handling data quality etc.
- Define and document project methodology, e.g. agile, scrum, Kanban
- Define and document development process, e.g. ETL (and tools), version control, development checklists, environments
- Define and document deployment process
- Define and document testing process
Last but not least implement all of the above.
Snowflake performance tuning and design
You need to understand various key Snowflake concepts to make good design decisions and tune Snowflake for performance
How does separation of compute from storage work?
What are micro-partitions and clustering keys?
What are performance implications of using the various loading techniques?
How does caching work?
What are column projections and predicate pushdown?
These are just some of the questions you need to consider when designing for performance
Have a look at some of the posts on our blog:
Migrating to Snowflake from Hadoop, Oracle, Teradata, SQL Server
- Did your Hadoop distribution over-promise and under-deliver. You are not alone.
- Are license costs for Teradata affecting the ROI of your data warehouse?
- Do you find it painful to reload your Redshift cluster every time you restart it? Do you have concurrency issues?
- Is your Oracle or SQL Server data warehouse bursting at the seams?
- ... or do you just want to take advantage of the many features that make Snowflake the data warehouse platform of the future.
There are many reasons for migrating from another data warehouse platform to Snowflake. Whatever your reasons, make sure to talk first to the Snowflake experts at Sonra.
Here are just some of the things you need to consider when migrating to Snowflake:
- How do I get the data from my on-premise OLTP systems into Snowflake?
- What are best practices for Snowflake virtual warehouses?
- What advantages do I get from separation of storage and compute?
- What about ETL? Do I need a new ETL tool? What are good practices?
- What are the differences between my current data warehouse platform and Snowflake
- What is the quickest way to migrate my data models? What data modeling tools support Snowflake?
- What are the best practices for performance? Hint: window functions are one component (link to posts on performance).
- How can I integrate Snowflake with my data science platform
Snowflake Data Lake Implementation
A data lake is a central hub for raw data from your data sources. It is similar to a staging area in a data warehouse. The difference being that the data lake has additional data consumers and caters for additional data types. Contrary to common perception you don't need Hadoop to build a data lake. Actually, we would advise you against this. Hadoop does not separate storage from compute. Whenever you increase storage you will also need to increase your compute capacity by adding additional nodes to your cluster. This can get expensive quickly.
Snowflake is perfect for building a data lake.
You only pay for storage of your data
With the data sharing feature it is easy to feed downstream consumers such as the data warehouse, sandbox environments, data preparation tools, and data science platforms.
With the virtual data warehouse feature you can create virtual sandboxes without the requirement to move around your data
You can load semi-structured data into Snowflake and query it in place
You can create external tables over data on S3 to query it directly