Modern Data Architectures
May 20, 2020

Data Visualisation in Modern Data Architectures (trying a webinar)

By Paul Laughlin

We’ve spoken before about how to run your own workshops online, but you may also want to attend others to keep learning, e.g. about Modern Data Architectures.

In a bid to experience for our readers whether the raft of webinars being advertised can replicate the events they replace – I signed up to attend one. Many working in the Data & Analytics space will have at some point attended Big Data LDN, as a one-stop show to see several vendors at once.

This is not my preferred type of event, but I noticed that in a series called “Modern Data Architectures” they had a webinar on Data Visualisation. I was also curious to watch this as it included a presenter from Looker, a tool that Ryan de Rooijen has recommended to me.

It soon became apparent that this event was more aimed at the BI audience than our normal readers, but I hope my takeaways are still useful.

Experience of a webinar replacing a live event

It’s fair to say this webinar was a mixed experience for me. There were some useful points shared, but quite a bit felt very passive and potentially boring. I can see why so many fall into just listening to webinars & getting on with other tasks in parallel (reducing their concentration on both).

So, let me try and share a balanced assessment with a mixture of praise & critique.

The overall approach was what has become the new style for webinars. An initial introduction from speakers & host, all of whom you see via video conference windows. Then as each presenter speaks this view is replaced with just watching a PowerPoint presentation or screen share of a demo.

Area for improvement

With talks lasting up to 20 minutes at a time, this is very passive for the audience, who have no way to interact apart from raising a question later. At this event, you also couldn’t see questions raised by others nor have any way to network with other attendees. Perhaps a missed opportunity?

A benefit of using webinars

On the positive side, if you are unable to watch it all live, then you are reassured that it has all been recorded and will be available for you to revisit at your leisure. I can see that is very convenient. However, as an educator, I regret this move away from live interactive learning.

The event was chaired by Andy Stead (from Big Data LDN), who calmly kept us all on time. The two presenters were Mike Ferguson (from Intelligent Business Strategies) and Ciaran Doyle (from Looker). They both shared points of interests, so let me summarise my takeaways…

Modern Data Architectures and Businesses’ BI needs

Mike kicked us off with a summary of both traditional and modern data architectures. At it’s simplest, these slides just showed that the data needs & thus infrastructures of businesses have got a lot more complex.

Not only are there many more sources & systems involved, but the variety of types of data has also expanded. Many organisations are dealing with diverse formats and some genuine Big Data (like streaming data from devices or timely clickstream data). Most also have the added complexity of hybrid data platform solutions (some in clouds & some on-premises).

Mike made the useful observation that most of the major BI ‘data visualisation‘ tools pre-date such added complexity. As a result, their ability to ingest or query data is designed for the days when that meant SQL queries of a data warehouse. This is an important point. I have long bemoaned that lack of a ‘layer of abstraction‘ to protect Tableau users from the performance-sapping mess that is most client’s data sources.

Critical factors for effective BI

He also shared some interesting research on the most critical factors for effective BI. This highlighted 3 top success factors:

  1. High-quality trusted data
  2. Shorter time to value (insights & action)
  3. Improved effectiveness of visualisations

I’m pleased to see that all three have been themes on this blog. Hopefully, that means our content is also helping the BI community.

Looker as a Data Visualisation tool for Modern Data Architectures?

Following Mike’s focus on architectures, AI & other strategies – the second half of this webinar was a talk & demo from Ciaran. He provided a useful introduction to the Californian tech start-up background of Looker. A popular tool that has reached the traditional stage of being snapped up by a larger tech incumbent (in their case Google for their Cloud offering).

Why the back end is better

Ciaran did a good job of explaining why this more recent tool is better adapted to the complex data architecture that it has grown up amongst. Looker uses a Virtual Data Platform that provides that layer of abstraction that I called for earlier.

This took me back decades, to my days of tweaking the ‘Universe‘ layer for Business Objects installations, this allows IT to tailor the data access to your own architecture. Potentially this should improve performance & protect against impacting other data platform users.

Room for improvement in Data Visualisation

If the back-end impressed me, I was less impressed by the front. It seems that every since Tableau appeared on the scene, BI tools now want to position themselves as Data Visualisation tools. From the demo that Ciaran gave of Looker, I think that is over-hyped. It does look like an easy to use BI tool. I’m sure it would help many BI analysts. But, it lacks both the range of potential chart types & the templates to guide analysts toward Data Viz best practice.

(aside from me) Take a look at Flourish

As I shared as my takeaway from #datavizlive in London earlier this year, the ecosystem of true Data Visualisation tools has increased massively. If you are interested in reviewing your toolset, I recommend looking at the summary I published of Lars Verspohl’s very helpful workshop.

Suffice to say that since that event, I am becoming a fan of Flourish (an easy to use Data Viz tool that guides the user towards best practice using templates). If you’d also like to look into that tool, I recommend watching this introductory tutorial from Data Viz guru Alberto Cairo:

Where to watch Modern Data Architecture webinars for yourself

All that said, I am still glad that I spared an hour to attend this webinar. It has increased my knowledge in this area and is available for you to review through this website (as well as links to others in the series):

Modern Data Architecture Series: Data Visualisation | BIG DATA LDN

Modern Data Architecture Series: Data Visualisation | BIG DATA LDN With multiple analytical systems now being implemented in organisations, a major challenge is how to empower end users not only to leverage these systems and be able to connect to data within them, but also to enable them to invoke analytical assets they create such as machine learning models and graph analytics.

Are webinars working for you, if so which?

Are you also giving more webinars a try? Perhaps you are even dipping your toe into the water of broadcasting some yourself? If so, please do share your recommendations in the comments section below or on social media.

Let’s keep sharing what we learn & engaging as a community online.