Episode 11: How Rackspace turned data liabilities into assets through cloud migration

In Cloud Consulting

Episode 11: How Rackspace turned data liabilities into assets through cloud migration - read the full article about cloud migration, Cloud Consulting and Data migration, Cloud infrastructure management from Google Cloud on Qualified.One
Google Cloud Tech
Youtube Blogger


Im Bruno Aziza.

And welcome to another episode of Data Journeys.

This is where we come to learn from data leaders, their dos, their donts, their best practices, and their exceptional results as theyre migrating to the cloud.

Today, Im really excited to have Juan, who is the Chief Data Officer at Rackspace.

Weve done a lot of work together, and hes got a great story to share with us.

Juan, thank you so much for your time today.

Welcome to the show.

Lets get started.

Tell us about yourself, and the company, and your journey.

JUAN RIOJAS: First of all, Bruno, thanks for having me on.

Im really excited about this.

I mean, this is one of the topics that I absolutely geek out about, Im fairly passionate about.

So my name is Juan Riojas.

Im the Chief Data Officer for Rackspace.

My responsibility is looking at data analytics internally for the company and really truly taking and elevating data as a strategic asset.

Im not sure if youre familiar with Rackspace, but if youre not, we are a global multicloud solutions provider.

We have over 140,000 customers.

We are in multiple geographies.

And we ultimately help enable transformations all around data services, multicloud, private cloud, security.

BRUNO AZIZA: And so now youve been at Rackspace for a couple of years.

Now, previously to this, you were at Informatica.

You also worked at Accenture.

So you have a really diverse perspective on this world of data analytics.

Youve been in the space for a long time.

So tell us a little bit, when you started at Rackspace, what were the use cases? What were the problems you were brought in to kind of solve or optimize for? JUAN RIOJAS: My background is what I think helps kind of look at data holistically.

You saw my first half of my career-- I called out Accenture, but my first half of my career was all in supply chain.

In supply chain, you always have to look at how all the dots connect.

Because everything has an effect.

For every action, theres a reaction.

So I had a little bit of consulting, I went into the consulting world.

It helps you look at problem-solving.

It helps you look and really refine how you look at opportunities.

I fast-forward now into what were doing at Rackspace, and I can certainly see how we wanted to go ahead and take our data seriously.

And when we started investing into that, I think one of the biggest tenets that we saw was really looking at how do we start driving better customer experience through NPS.

That would be the result. And then secondly, in this quantum world where everythings connected and its this subscription-based economy, its how do we mitigate our churn.

If we can do those things and you can get it right, thats really what sets you apart.

BRUNO AZIZA: And they created the role for you.

There was no chief data officer at Rackspace before.

So can we talk more about this? One, the customer experience and churn, how is that measured and how do you construct an effective data infrastructure around these two business objectives? JUAN RIOJAS: This data office did not exist.

What you had these multiple organizations that were decentralized.

And we brought them all together.

And what we had in churn was, for example, there was 14 different versions of how we measure churn.

On NPS, there was multiple different versions of how we look at NPS.

And on those two things, how data and the infrastructure helped us get to a significant improvement was by bringing the data together.

At Rackspace, when I first came in, we had 70 ODSes and four data warehouses, a data lake.

It really wasnt being manager-governed to the best of its ability.

And by bringing all the data together, we were effectively being able to go ahead and, one, harmonize the data, get it in near real time.

We were able to go in and standardize on a definition.

So we had integrity around our metadata.

And then third is we exposed it in a way that was easy to go and understand, to look.

And when you measure, you improve.

So what we saw was, quickly, when we brought the infrastructure together on GCP and we were using, for example, ingestion with-- we started with Alooma, and then it went to Airflow, and we went to Dataflow, we started seeing how we could actually stream this information.

So going from trying to go ahead and talk about the integrity of our information, we were able to go ahead and get this information near real time.

And more importantly, we start seeing improvements.

As an example, NPS, we grew 10x of where it was.

We continue to see improvements.

That fundamentally shaped our entire predictive analytics strategy to start looking at churn not in a reactive state, but a proactive state.

By incorporating all the data and having it in one place, we were able to accelerate our journey to using TensorFlow to start using decision trees, to start bringing the modern data together to start looking at customer experience and really translate that to operational sense, whereas we are looking at how we can actually go and drive improvements in churn by looking at 40 different variables.

BRUNO AZIZA: So lets talk a little bit about these business results.

You started talking about the definitions and the centralization of data and so forth.

So did you go from a very decentralized way of looking at data to a centralized data warehouse that was getting feeds into in real time? And if thats the case, then how much time did that take you? And what changes did you make? Did you have to hire new people? How did that all work? JUAN RIOJAS: Yeah, we did.

I mean, I think one of the biggest challenges is I think data was a liability for us at the beginning.

So when I first came in and we were averaging about 80 defects in production a month-- a month-- which is unprecedented.

But because we had no standards, there was no quality, even though we had data warehouses, they were built on on-prem environments and they were fairly archaic.

The code hadnt been updated in years.

The data model was fairly antique.

So by bringing it all together-- and when I say all together, we effectively migrated a petabyte of information, the 70 ODSes, the four data warehouses, the one data lake that we had on Hadoop.

We structured the team, we all learned how to ramp up on GCP, we all learned how to use SQL and Python, and we effectively migrated everything within six months, which for me was fairly impressive, because that really started driving the business outcomes.

And typically what you see in transformation, especially in data transformation programs or modernization efforts, if youre not quickly showing value, the energy starts fizzling out.

So one of the biggest tenets was really planning effectively, getting everything migrated, getting everything migrated with quality, to the point that every single dashboard that comes from us, every single data source, has our seal of approval.

Kind of like-- think of this like an iso seal.

Were called GDO, so Global Data Office.

And every dashboard has our seal in it.

If it doesnt have our seal, its really not going to be-- I wouldnt say trusted, but its not going to be representative of what we have.

Our seal, we use a traditional supply chain concept of a perfect order.

And we converted that to data.

If we have a definition for the metric, if its defined, if its secure that its encrypted at rest and in transit, and that we check for completeness or accuracy, integrity of the information, if it has those three things, then it receives our seal.

And thats what we expose.

So there is no ambiguity of how you use that information.

BRUNO AZIZA: So theres a lot here.

The dos and donts is the next section were going to get to.

Before we get to that, I want to talk about the concept of certification.

If I want to certify data the way youre doing it, Juan, where do I start? Do I hire new people to do it? Is this automated through algorithms? How does this work? JUAN RIOJAS: Its not automated yet, but it is a manual process where we take a simple approach of making sure that we have the three tenets of quality, that we have security, and that we have a definition of governance around our metrics.

It is manual right now.

Whereas we dont have to invest in it.

Its really the same team members.

Our BI team as an example, all the dashboards, they ensure that those three tenets are clearly defined, theyre all aligned.

And then once its done, then we will put our seal on the dashboard as an example.

Or when we pass data are, our team will go and share that specific seal.

It is manual right now.

But we do want to get to the automation piece.

And I firmly believe that that was the catalyst for us to go from less than 200 users of our information, our data, to over 2,000 concurrent users.

So much so that I think the average user of even our dashboards, before they used to go and look at it in less than a minute.

Right now, the average person looks at it for 26 minutes, meaning that they are going in and theyre going deep in thought or going deeper.

And now were using those analytics to go ahead and help improve that experience and improve more insights to the organization.

BRUNO AZIZA: So youve got a full cycle here.

Not only are you increasing the number of people adopting, but youre increasing the attention.

And all of that is based on the seal of trust, in a way, that your team is applying to this data based on key principles.

Lets talk about some other dos before we get to the donts.

What is the one thing that people listening to us here that are trying to follow you in a similar journey, an exceptional journey, what should they absolutely do? JUAN RIOJAS: I highly encourage you all to really look into planning.

We created something fairly unique in our organization, which is we created a product office.

We start with three tenets-- data as a service, data as a platform, and data management as a service.

So data management as a service, our governance, is a platform.

And then our infrastructure data as a service is how we enable, how we consume our information, how we keep on driving the insights.

So investing into an organization like that is going to pay dividends.

Because at the end of the day, you want to take out all the requirements, all the business partnership, all the technical specifications so your team can really focus on what they can do right.

It is on really building the models, taking advantage of the next-gen analytics, really bringing together optimization strategies.

And I think that is absolutely pivotal for what I would consider doing.

That and we incorporate also scaled agile.

So being able to go in and have a defined methodology, and then coupling that with the planning.

I think that, for me, was a recipe for success.

BRUNO AZIZA: So both process and planning and doing that early as you move to your transformation journey.

Now, what about the opposite of that? What is the mistake that most people do, or that you see, or maybe that youve done yourself, that you want to make sure everyones avoiding.

JUAN RIOJAS: I think the last thing is I think everybody has great ideas.

Theres always going to be edge cases.

Sometimes those edge cases, you may just want to go and leave them to the right.

Put them into a parking lot.

Because when you spend a significant amount of time to go and solve for that less than 1% of the issues, thats when it just drives frustration, it drives delays, and I would rather go ahead and focus on getting the bigger pieces right, start driving value, and then look upstream as to why those exceptions are happening.

I think thats something that I would recommend do not go do right now.

BRUNO AZIZA: Juan, thank you so much for your time today.

We learned a lot.

We learned about data certification and trust.

We learned about the importance of planning and having a process to migrate to the cloud.

And, of course, we learned not to optimize for exceptions.

This was an exceptional data journey.

I hope people are going to reach out to you to learn more about your journey.

Thank you again for spending the time with us today.

If you want to find out more about stories just like this, make sure the click, down below here, to the link of many more stories like this.

Juan, I want to thank you for your time.

I hope people are going to reach out to you and learn from you.

Until next time, Im Bruno Aziza.


Google Cloud: Episode 11: How Rackspace turned data liabilities into assets through cloud migration - Cloud Consulting