2. What is ?
● Social Network for meeting people near you
● Over 300M users in 190 countries
● 200 employees based in London and Moscow
● Over 100M downloads on Android
● Also on iOS, Windows Phone, Web and Mobile Web
● Very Agile working environment
3. ● We needed a proper BI tool - among 14 candidates
● Our data volumes - user level data
● Environment - Linux, Database, SSO
● Technical users with high expectations
Why ?
4. ● Hard to set up in our environment
● No real dimensional model
● Data/ETL Team had to prepare data for us
● Time to onboard users and earn their trust
vs
5. What does with
now?
● Fancy Dashboards around the office
● Data Discovery tools
● Analysis delivered by email
● Self Service Reports
● Weekly releases
90+ Users in Finance, Billing, Marketing, Developers, User
Ops, Founders
8. Badoo’s Database
EXASOL is an Massive Parallel Processing (MPP) database.
It is an in memory columnar database.
● 8 Nodes (plus 1 spare) with 5.6 TB of RAM
● ~100 TB of Raw Data - ~30 TB of Data on Disk
● Each node has 8 TB of Disk, in RAID 2
and redundancy factor = 2
9. and
Query Generation Time: 0:00:00.13
Total Elapsed Time in Query Engine: 0:18:36.68
Sum of Query Execution Time: 0:16:08.46
Sum of Data Fetching and Processing Time: 0:01:03.73
Sum of Data Transfer from Datasource(s) Time: 0:00:57.93
Sum of Analytical Processing Time: 0:00:00.00
Sum of Other Processing Time: 0:01:24.49
Sum of Cube Publish Time 0:19:06.37
Number of Rows Returned: 5759450
Number of Columns Returned: 38
Number of Temp Tables: 0
Total Number of Passes: 15
Number of Datasource Query Passes: 15
Number of Analytical Query Passes: 0
10. Query Improvements
● Use real tables
● Use parallelization
● Use the Pre/Post Processing statements
and
11. and
Query Generation Time: 0:00:00.13
Total Elapsed Time in Query Engine*: 0:11:18.82
Sum of Query Execution Time: 0:18:30.86
Sum of Data Fetching and Processing Time: 0:01:04.82
Sum of Data Transfer from Datasource(s) Time: 0:00:59.37
Sum of Analytical Processing Time: 0:00:00.00
Sum of Other Processing Time: 0:01:55.12
* This report has some passes that have been executed in parallel.
Individual time components may not add up to Total Elapsed Time in Query
Engine.
Sum of Cube Publish Time 0:11:28.63
Number of Rows Returned: 5759450
Number of Columns Returned: 38
Number of Temp Tables: 17
Total Number of Passes: 49
Number of Datasource Query Passes: 49
12. and
Query Generation Time: 0:00:00.09
Total Elapsed Time in Query Engine: 0:02:43.29
Sum of Query Execution Time: 0:00:53.08
Sum of Data Fetching and Processing Time: 0:00:52.09
Sum of Data Transfer from Datasource(s) Time: 0:00:47.83
Sum of Analytical Processing Time: 0:00:00.00
Sum of Other Processing Time: 0:00:58.10
Sum of Template Calculate Time 0:00:00.00
Sum of AE Data Persisting Time 0:00:00.46
Sum of Cube Publish Time 0:02:56.36
Number of Rows Returned: 4702678
Number of Columns Returned: 38
Number of Temp Tables: 12
21. Enable Your Users with Transaction Services
● Agile environment
● New analysis have an assessment period
● People just like to play with data
This was just bad
Time consuming
Self esteem problems
We would end up hating our users
22. Enable Your Users with Transaction Services
“We have a Coefficient that we would like to use in our
calculation, this can be different for Campaign Media Source,
Country, and Platform...”
200+ Media Sources
254 Countries
12 Platforms
200+ x 254 x 12 = 609600!
23. Enable Your Users with Transaction Services
We had to convince them to have a go with
Transaction Services!
28. Don’t reinvent the wheel: use MicroStrategy
Problem: Deliver a csv file to an external location.
Proposed Solution:
❏ Generate the data
❏ Put the data on a local drive
❏ Create a tool to copy it remotely
32. Few things we do with Command Manager
● Cube Refresh
● Start Schedules
● Manage Our Users
● Configure Database Connections
Save Time with Command Manager
33.
34. My two cents about Command Manager:
● Get familiar with it
● Try to script repetitive tasks
● Integrate it with other tools
Save Time with Command Manager