Snowplow is an open-source analytics platform that can be installed on AWS. They handle the event data pipeline, collect data from multiple sources, enrich & sanitize and send the events to your data-warehouse. We’ve learned a lot from Snowplow, as their workflow is quite nice and flexible. The main problem is it’s still not easy to setup. As with many AWS services they provide you the tools separately, and you’re required to learn their nomenclature and API calls to glue them together and build your own solution on top of Snowplow.
Both Snowplow and Rakam provide you an event data pipeline solution that allows you collect data from multiple sources, enrich and sanitize the events with event mappers and store the data in our data-warehouse. Both solutions allow you to select your data-warehouse of choice, with Snowplow supporting Redshift, Postgresql and Elasticsearch and Rakam supporting Postgresql and PrestoDB.
Postgresql is flexible, real-time and fast for small and medium size data volumes but unfortunately not horizontally scalable and the default row-oriented storage format has too much overhead in big-data. For larger data volumes, we have PrestoDB deployment type.
Unfortunately most of the data-warehouse solutions such as PrestoDB and Redshift are not real-time. Therefore, we forked PrestoDB and implemented a near real-time data warehouse on top of it. Since ad-hoc queries are not enough for dashboards, we also implemented stream processing inside the database with continuous queries, just write the SQL query and create the continuous query background job. Rakam will process the data at ingestion time so you don’t need to process billions of data when visiting your dashboards. Since we have full control on our data-warehouse, we also allow you to mount your remote database schemas as virtual tables, mount your remote files (CSV, AVRO etc.) as virtual tables so that you get enough flexibility to create your custom analytics service easily.
A caveat with Redshift is it’s fully managed by AWS. You need to load the data to it when you want to analyze and even if you usually run queries on small part of the data-set (last month, last day etc.), Redshift still charges you for all the data you have. With PrestoDB, our customers usually start more nodes when they need to run complex queries on larger data-sets and terminate them when they don’t need. It affects your bills at the end of the month because you will be allocating the real resource you consume on the cloud.
To make it easy to interpret your data before doing any additional work we include a BI that allows you to create user interface for your analytics service. It’s not exactly the same as other BI solutions such as Tableau, with a focus on creating an UI for analytics service similar to Google Analytics or other specialized analytics solutions. The end-users should be able to use the interface without any complexity. That said just like with Snowplow, if you already have a BI solution, you can pipe data without pause via the Rakam API.At Rakam we believe that a full-stack analytics service should be available with one click. Let’s connect about your big data project!