Welcome new joiner, Francesco Puppini

Shifting to cloud-first analytics is the next step in the evolution of BI.

At Astrato, we want to provide data teams with a modern BI and analytics platform that unlocks all the agility and scalability of the cloud. And we need the right technical expertise on board to continually drive innovation.

That’s why we’re excited to welcome Francesco Puppini, our new Astrato R&D Engineer! You may have heard of his groundbreaking Unified Star Schema.

 

We asked Francesco a few questions about his mission at Astrato

1. What was your role before Astrato?

I am a Business Intelligence and Analytics specialist. There are many software solutions in this domain, and I have spent quite a lot of time with many of them, studying in-depth how they work in the back end.

When data is easy, everything is easy. But when data is challenging, different products behave in different ways.

I specialized in understanding how they managed to solve a number of specific problems. I once raised a case with the technical support of Tableau, back in 2015. They solved it in 2020: five years later! My request was simply to visualize my data. Business Intelligence can seem very tricky sometimes! (😉 But we know it doesn’t have to be.)

2. What is your role at Astrato?

Now Vizlib is building Astrato, and I am proud to be part of this journey. I came here to share my experience, and bring the best of what I learned in the past years.

I know how problems were tackled by QlikView, Qlik Sense, Tableau, Power BI, Looker, Thoughtspot, Domo, Spotfire, GoodData, Sisense, Incorta, Clearstory, Business Objects, Cognos, Brio, Hyperion, and more.

If some of these names are not familiar to you, that’s probably because they don’t exist anymore! In Clearstory, they had the idea to match data by “affinity”, so I saw customers associated with orders based on their birthday.

This is wrong!

The tool no longer exists, of course. I am here to help build a solid bridge between Astrato and the data sources, leveraging the existing best practices, combined with innovative ideas.

3. So what inspired you to build the Unified Star Schema (USS)?

I saw a gap in the literature available on data.

We live in a world that is full of software solutions, but in some cases, we struggle to frame the problems we’re trying to solve. And this goes beyond the domain of data, I am afraid.

A solution is impossible if we don’t know what problem we have.

So, I tried to rethink the basics of SQL and Data Modeling. I started by categorizing the common challenges faced by data practitioners every day. I gave names and definitions to the most common “data traps”. These names already existed, but I gave some simpler definitions. I suggested a methodology to easily detect the traps, and finally, I indicated a way to prevent them.

All this is explained in my book.

4. I’m not a data modeling expert… could you explain the concept of USS in simple terms?

Let’s explain it from the point of view of a business user who needs some information. When you want to analyze your Sales, you can find your answers in the “Sales dashboard”. When you want to analyze your Purchases (materials, expenses, etc..), if you are lucky, there is a Purchases dashboard you can use.

Behind each dashboard there is usually a star schema: a structure of tables centered on a topic. Why star? Well, with a bit of imagination, the shape of this structure is similar to a star. Your main topic is “in the middle”, and then a series of extra details are all around it, giving you the freedom to analyze your revenue by clients, by products, by suppliers, and so on.

But what happens when a business user wants to analyze Sales and Purchases at the same time?

This is a legitimate request because profit is way more important than revenue on its own! In most cases, for an analysis on profits, the business users need to make a new request to their IT department, and this can take several weeks. In some cases, they can’t afford to wait and end up downloading both data sets to Excel, and merging the information manually.

This is NOT what I call a modern BI solution! So, here is where the USS comes into play: I have found a way to harmonize all these pieces of information into one single “Unified Star Schema” – a data structure where all the star schemas are seamlessly integrated.

5. How do you think this model helps businesses modernize their BI?

Self-service.

The USS lays the foundation for a very powerful self-service data platform. Let’s make an analogy. When you want to make a phone call today, you just do it yourself.

But if you think back to the 1920s, you had to rely on switchboard operators to physically connect cables to the right terminal for you. With Business Intelligence we are still in that old phase, unfortunately. Business users today have very poor access to the information they need because they need an “operator” – someone who develops a script or a dashboard for them.

Self-Service analytics, today, does not exist for business users. A modern BI solution should facilitate and democratize access to information.

Please note: this does not replace the work of the developers, but rather gives them more free time to deal with more advanced tasks.

6. Tell us more about how your innovative thinking (and any new models you’re working on) will help Astrato support modern data teams?

Every analytics solution claims to be a Self-Service solution. But have you ever seen a Manager opening a blank Tableau document and connecting tables?

This will never happen! Business Objects 30 years ago tried to propose the concept of “pre-connected tables” (they called it “Universe”). I designed many Universes in the past, and I soon realized that “Universes were not Universal”.

It was impossible to pre-connect the tables in a universal way because the way we connect tables always depends on the business requirements. But then I realized that we need more than this! Business users want to be able to query a database dynamically, and without the help of the “middle man” who writes code for them.

This is not a small challenge.

My approach uses a new principle: A relationship between two tables is always DIRECTIONAL. When we join two tables, one table is always the hunter, and the other is always the prey. If we do not take this into account, then we risk “breaking the numbers”.

Imagine you send an invoice to your client with an amount of 100 on the invoice. If you join the table of invoices with the table of email addresses, and your client has 3 email addresses, the resulting SQL query will show an amount of 300, which is incorrect. It does not matter if we use INNER, LEFT, RIGHT, FULL OUTER JOIN, or even a DISTINCT statement: the result will be incorrect in each case. The key principle is that “certain combinations of tables cannot be joined together”.

If we take into account the direction of a join, we are able to prevent this problem. Based on the hunter-prey logic, Astrato will feature a brand new way of writing SQL.

7. What is your vision for taking Astrato to the next level?

Thinking about the customer experience.

Most books about data are written for technicians or theorists. My vision is to help build a smooth customer experience, with a particular focus on business users.

Accessing and exploring data should be as simple as adding items to a shopping cart. When you buy books online, have you ever experienced any constraint in your choices? I don’t think so!

If you add “Excel for Dummies” to your cart, nothing stops you from adding “Pride and Prejudice” next! (This may give a hard time to the suggestions algorithm, but that’s not your problem!).

With data today, things are different. In SAP Business Objects, when you pick two incompatible dimensions, it is quite common to receive the error: “This will generate a Cartesian Product”, or “Incompatible combination of objects”. Most business users do not even know what this means, and they definitely don’t know how to deal with it.

In other cases, there is no error message at all, but the results contain errors. Everyone has experienced “weird numbers” in their reports. And often, no one can really explain why this was happening.

The Unified Star Schema eliminates this problem at the root.

With the USS, numbers are always matching the data source, with no exception. A business user may ask a “non-coherent question”, selecting business elements that do not have any correlation.

For example, a query asking “Stock Amount by Client Gender” does not make any sense, because the products in stock are never associated with clients. In this case, the USS will simply display the results on different rows. No error messages. No Cartesian products. No inflated numbers. No long waiting. No expensive cloud bills. Astrato will deliver the result immediately.

In addition, it will display a warning to the business users, guiding them on how to build a more sensible business question. This is actually simpler than it sounds. It will not need any Artificial Intelligence algorithm. It will be based on that small piece of SQL theory that the data literature left behind for many years: the direction of a join.

This is my vision.

8. What do you love most about your new team?

Kindness. I have worked in many large corporations, and this is my first startup. It’s a completely different space. The first thing I noticed is the kindness and positive attitude that everyone has. Startups may have tight deadlines, but kindness does not slow down productivity. It’s actually the opposite! 😄

9. What motivates you to work with your team?

Open-mindedness. I am motivated by how open-minded the team is. I am here to bring new ideas. Without that magic ingredient, the collaboration would not work. And I must do the same – my ideas can be improved, expanded, or even questioned. I always hope to find someone who has better ideas. It’s the whole point of working in a team!

🙏 Thank you for sharing your thoughts with us, Francesco. Your passion and skills will help shape Astrato and ensure BI is simple and accessible for all users!