Looker on GCP Meetup (7th May 2020) - Q&A

May 14, 2020 | in Looker
5 mins read

Looker+GCP_QA

Q&A from Looker on GCP Meetup (7th May 2020)

Last Thursday's meetup at the Google Cloud Data User Group had some great energy with a variety of interesting questions. Thank you to everyone who attended and for asking. Unfortunately we ran out of time during the session. So as to leave no stone unturned, we've anwered all the questions posed during the  Q&A below.

If you need any clarification or follow up on the answers then don't hesitate to drop an email to looker<at>agilexi.com

Questions

  1. How integrated/bound to Google is Looker? Do you need to grant permission to Looker to be able to use it? What is the cost of using Looker? » answer
  2. In Gartner's Magic Quadrant for 2020, Looker is in Challengers quadrant. What Gap/step is required to move it to Leaders? » answer
  3. Do the reports respect permissions on the backing tables/views in the database, or does Looker have a separate authorization mechanism? » answer
  4. Where does Data Studio fit in around Looker? » answer
  5. Are there customers who have Looker as part of their data science environments or is it primarily for BI? » answer
  6. How much of that drill down feature is automated in the product or required to be defined for it to be enabled? » answer
  7. How are we accessing weather data in this example? Are we accessing weather data from BQ or from other website? » answer
  8. Is there a PagerDuty integration option for alerts? » answer
  9. Is there a way to host Looker code in Github? we would like to be able to manage the code similar to all of our other code? » answer
  10. Can you save your query and run it on a daily basis to generate a report on certain periods of time? » answer
  11. Can we extend looker model as a building block of another model? » answer
  12. All the examples you have showed which use AppEngine, Cloud Functions, Tensorflow/ML Engine. Is it possible to use Looker with the Azure or AWS equivalents? » answer
  13. If ERP data is stored in database. If customer wants to user looker. Does looker provide plug and play without any development » answer

    New call-to-action

Answers

1.   How integrated/bound to Google is Looker? Do you need to grant permission to Looker to be able to use it? What is the cost of using Looker?
Looker is cloud agnostic therefore you do not need to be on GCP to use Looker. You will need to grant Looker permission to query your database. For GCP you can do this either through OAuth or via Service Accounts that you create within the GCP console. The cost of using Looker is custom and will depend on the number of users within your organization, data connections and scaling.
There are volume discounts and the like. Looker does offer a free trial.
For a quote feel free to speak to us  or contact Looker directly.
↑top

2.   In Gartner's Magic Quadrant for 2020, Looker is in Challengers quadrant. What Gap/step is required to move it to Leaders?

As the Google acquisition was not complete at the time authoring the 2020 MQ, Gartner has not considered Google in its assessment of Looker. It will be interesting to see how Looker’s positioning shifts when viewed a part of Google’s wider offering - the effect can only be positive (presumably!). On the cautionary side, two  of the Gartner's concerns related to the narrowness of vision and lack of Geographical presence. Both of these should also be abundantly addressed by becoming a part of Google Cloud.

Gartner's third caution related to ‘Power user skill requirement’ for modelling data. All the leading BI vendors, i.e. PowerBI, Tableau etc, have comparably technical modelling languages (i.e. DAX, LoD expressions etc). However they also have a visually driven, user friendly side that allow non-technical data practitioners to quickly explore and model the data through a UI. This is missing in Looker at the moment and could do with improvement.

Gartner also mentions a lack of data preparation capability. However Looker’s approach has been to delegate physical preparation and storage to the data sources and focus on the data modelling (which implicitly does provide opportunity to prepare the meaning / definition of dimensions / measures).
↑top

3.   Do the reports respect permissions on the backing tables/views in the database, or does Looker have a separate authorization mechanism?

Either of the two can be used. Looker can provide the security by applying authorization via user management and dynamically modifying the queries based on the user viewing the data. The alternative is to leverage the user permissions coming from the database.
↑top

4.   Where does Data Studio fit in around Looker?

Looker is enterprise scale i.e. if you need scalability, data security, governance and multi region availability, then Looker is the right tool. Looker also allows self service and independence for stakeholders utilising the tool. It is a full blown data platform that can be used to power custom applications and/or integrate with other existing applications.

Data Studio, on the other hand, is a great tool for someone who might be a bit technical and wants to answer questions from their data and share with a few other people. Data studio allows you to upload CSV files and start exploring or append to other datasets. It does not need to have a database connection.
↑top

5.   Are there customers who have Looker as part of their data science environments or is it primarily for BI

Yes there are customers that use Looker for their data science workflows, leveraging ML and Tensorflow via BigQuery directly or by providing governed cleansed data into their ML model via the Looker API.
↑top

6.   How much of that drill down feature is automated in the product or required to be defined for it to be enabled

Looker allows you to drill down by specifying drill_fields when defining your dimensions or measures. It is not there by default. You can define the level of drilling or create your own custom drill path in the LookML.
↑top

7.   How are we accessing weather data in this example? Are we accessing weather data from BQ or from other website?

The data is from BigQuery, BigQuery offers a series of public datasets that can easily be combined with your data. Looker connects to your database for all its modelling. Looker also has some data blocks available.
↑top

8.   Is there a PagerDuty integration option for alerts?
Not as a default integration but as the action hub is open, you are able to add your own PagerDuty integration in Looker and it will look and work as natively as the other integrations.

↑top

9.   Is there a way to host Looker code in Github? we would like to be able to manage the code similar to all of our other code?

Yes, LookML uses Git for source code version control. All the code that generates is managed in your Git of choice. Read here for more details.
↑top

10.  Can you save your query and run it on a daily basis to generate a report on certain periods of time?

You can save a Look or a Dashboard that you generated, which retains the source query / queries. You can specify a schedule at which to run the Look or Dashboard and finally the delivery method, destination, and format.
↑top

11.   Can we extend looker model as a building block of another model?

You can extend views and explores, include model files and import whole projects. See the extend parameter and the local project import.
↑top

12.   All the examples you have showed which use AppEngine, Cloud Functions, Tensorflow/ML Engine. Is it possible to use Looker with the Azure or AWS equivalents?

Looker is cloud agnostic and you can indeed use the equivalent functions from other providers.
↑top

13.   If ERP data is stored in database. If customer wants to user looker. Does looker provide plug and play without any development?

Looker will generate an initial model for the database it is connected to. THe model will contain a view file for each table in the database and in those a dimension for each column. From that point on, the data is accessible and could be used with custom fields, however to provide a rich governed experience to your users, further development of the model is required. Realistically some level of development will always be necessary.
↑top