What we learned from building custom analytics over Amazon Redshift

Re:What we learned from building custom analytics over Amazo... by itamarwe on Monday November 30, 2015 @03:14PM (#10651450)

Yes, exactly. You'll push events from Alooma's Android SDK and they'll land in your Redshift in seconds. You can have a look at the documentation here: https://www.alooma.com/docs/integration/android-sdk

[ Parent | on HN ]

Re:What we learned from building custom analytics over Amazo... by buremba on Monday November 30, 2015 @05:46PM (#10652457)

I couldn't really understand how you replaced Mixpanel and other analytics services with Redshift. AFAIK Redshift is optimized for bulk inserts so the micro-batch approach (a few seconds) is not is not optimal and in fact quite expensive. Also the data in Mixpanel is not structured and if you're using JSON in Redshift, you can't take advantage of the columnar storage format which is actually the main point of using Redshift. Do you care to explain how you solve these issues?

[ Parent | on HN ]

Re:What we learned from building custom analytics over Amazo... by yairwein on Monday November 30, 2015 @11:37PM (#10653658)

Hi, Alooma uses batch size appropriate to the data volume, the bigger the data volume is, the bigger the batch size will be.

Alooma not loading into a JSON column, but provides mapping UI and transformation layer to in order structure and normalize the data into different appropriate columns, thus leverages the columnar storage properties.

FD: I am Alooma's CTO

[ Parent | on HN ]

Re:What we learned from building custom analytics over Amazo... by sandGorgon on Monday November 30, 2015 @03:18PM (#10651473)

*mAPI = AloomaAPI.getInstance(context, “<Company_Name>.alooma.io”)

no api key ?

this looks interesting - but it would be nice to be able to get started immediately without "request a demo".

[ Parent | on HN ]