Data Governance as a Service – Steve Fazio, Talend

This is texturing TV. Hey guys. Thanks for the throw.

We're here with Steve Fazio. Who is the product manager for data catalog at Talent? We're gonna be talking about all things related to data governance Steve.

Welcome the show. Thank you. Thanks for having me.

Appreciate it. Some people would argue that data governance is yet another example of a long line of it oxymorons and that there isn't a whole lot of governance going on in the world. So what is the current state of data governance and How do we kind of clean it up?

Because it is something of a mess. With a giant broom then that broom is gonna be called. I think the newest term is Degas, right?

And you know, so I kind of laugh and I see data governance as a service. That's a weakest. I think to your point.

We were we're the the required acronyms has dipped to an unacceptable position. So they had to add a new one called Data governance as a service now. So we have to learn that one and I think some of the the issue what you described is it's a mess sometimes as people use the word data governance and and they in certain organizations and they don't know what they're trying to accomplish right?

What are they trying to accomplish? They have a clear business directive. Is it compliance?

Is it a business analytics thing? So I think the data governance as a service is going to help, you know be that big broom that comes in and helps, you know, whether it be physical Services of data governance and and meditate experts. Let's say Off-site or offshore helping out organization or physical automation, you know more and more if you see more and more technology and more solutions is really focusing automation.

and augmentation what exactly is data governance as a service because people in their minds would Envision some sort of, you know, Cloud platform that I don't have to go and stand up all the infrastructure and drive this but what exactly is a governing and how is it working? Right? So it's a it's I've seen it in two different ways.

Right one. The services is basically services that are called. So in the past maybe you had data lineage right with something that you had to go into catalog you had to go look at and you had to set up and to do I see more that lineage being a service where you tell the service.

Hey, I want to I have these two these two endpoints and that service has access to all of our metadata and that service generates lineage and like lineage can be returned to wherever it's it's requested from rather than someone having to physically going to a catalog or or lineage specific tool to go look at it and it gets broader, you know things like stewardship data stewardship can Be service can be augmented and automated like data profiling data classifications and things like that could be more automated and be provided as more of a cloud-based service rather than somebody at a computer and doing that themselves. So does that mean we're getting better at leveraging the metadata that we're generating on all these platforms because that's always been something of a North star but very few people ever got there. Right good question the we are in general.

I always say as as an industry and and I see more more customers being able to Wrangle metadata. It's metadators the oil right if every company has so much of it, but you just don't know where you don't know where to drill some times with proper augmentation of the documentation of metadata to make it more readily available. Now, it's becoming more easier to actually use the meditative.

I don't think in my opinion we had much trouble pulling data and pulling metadata and there's always Bridges and things like that that are meditative ingest. Thank our problem in general when I see a customers or they don't know what to do with it, or they don't know where it is. So like you said, they need that Northstar they need a place to go.

Be able to understand what's there and with the advancements in things like data classification being automated being able to expose the metadata to the right people at the right time makes that oil. It's like the it's a map right now. You can see where that oil is Where to drill and where to get the most value out of that metadata.

How does that help bridge the divide that is always existed between business and it when it comes to data management treatment matter is it folks don't have a lot of insight into whether or not a particular piece of data is more valuable than the next they're basically collecting and storing it and they're in charge of the stewardship of the data, but the business folks know what data matters or should matter and It's always been an issue in terms of figuring out how to give each of those squads. They're right insights that they need. So, how are this all going to get bridged as we go for it?

So typically most when you talk about business, right they they are more in terms and Concepts and things like that's like a gloss you see like a business glossary of terms the connection between the terms and the physical metadata. That's the biggest thing and that's I where I really see automation becoming more and more powerful. So as you bring in more metadata that data can be classified and have terms attached to it so that the business can understand exactly what a piece of metadata is how they can use it and how it's related to all the other different things that they have in there in our ecosystem.

Where does AI fit in that equation that are we going to use that to attach the metadata or do the classification? I mean how smart and smart get well, it's as smart as the stuff you give it. It's a smart as the I guess the more you do the better you get at it as for profiling classification, for example, so machine learning especially for metadata data classifications.

So if you have let's say a domain that you're trying to identifying classifying there's customer information on there is pi information, you can use rather than just using the data itself. You can use pieces of metadata. The the machine learning can use an AI can use different metadata to help classify that data.

So things more like not just call them names and frequencies and and content and quality structure get started using things like relationships and glossary terms. So if a piece of data may have a term defined I can look and see other pieces of data other columns or tables or things that also are defined by that same term and that can help with the data classification. Right?

So it helps automate and and the more and more and more that that process has to pull from the more metadata that has the act more accurate is going to get therefore kind of reducing that that human intervention of data classification. Does this have cyber security implications? Because it seems like there's a lot of effort to secure data these days, but we don't even know what data is where so that makes it kind of difficult.

So what's the relationship right? That's the first step. So whenever I whenever I advise customers and what to do the first thing you have to understand is what you can what's available.

Right? And is that data? Is there Pi in there?

Is there anything sensitive is there anything that that we don't want to put out to the Enterprise. So things like sensitivity labeling automating sensitivity labeling so that when a piece of data comes in it could be automatically labeled as confidential top see identifying things from asking for obvious skating or for encryption, right? Those things need to happen at the time that the metadata is Injustice.

You understand exactly. Once that data gets into that catalog now, it could be with those those rules and policies one thing that's that's that really controls all this and things like policies, right? So as you create new policies as an organization those need to be actionable and those policies like you said, what's our policy on data privacy does need to be defined and as new data comes in data that meets that that policy requirement that needs to be attached also at the same time so that personal information and privacy is kept throughout the Enterprise.

As we go along who's in charge of this we see in some organizations. There's a chief data officer. They may have the actual title of the role but in the absence of that who takes the lead or what are you seeing in different organizations?

I've seen it all over the map, you know to the size of the organization dictates. Sometimes the maturity level of that organization. Sometimes it's the CDO office, right?

They're the ones that are the policy makers So when you say in charge, I I always say that folks that are making the policy. It's like government right you have we have police we have government without the government giving the police. The laws the police have nothing to enforce same thing in organizations.

They need to have that organized that governing body that's actually putting forth those policies. And typically it is the the CDO office. Sometimes it's in the line of business.

So that line of business is responsible for their own areas and they're the ones in charge for their own domain. there's been a lot of folks that I've seen that have kind of combined roles now data stewards and data Architects are kind of sometimes they're the same they take the same role but it really does depend on the size of the organization and the the number of domains they have As we go along here, are we in danger of being overwhelmed by the amount of data that we're generating. It's not only the volume but the speed at which it's coming across and we reach some sort of constitutional.

Crisis Breaking Point. Um, if it's constitutional say I'd say we got you know, my I'll put it nicely. We have 10 pounds of baloney at a four pound bag right now as more more and more data the ease of getting data and the volume of veracity of data, especially as you know, probably during covid.

I see during that whole time there were so much more digital activity going on everything from streaming to shopping to Home Health Care all of that generated just exponentially, I don't know the number exactly but exponential amounts of new data some of the very valuable. I think again, it's these we have an oil field a tremendous amount underneath the ground, but we just don't know where to drill and it is getting to a point to where if and the manual processes. It's just too much too much to do sit for people to sit down at a catalog or at a metadata Management console store pointing to all these things to do it.

The the services and the automation will help I think. Perhaps Wrangle in some of that overflow of data that we have now, but I do believe we're at a point where we're already a point where I think organizations have more data than they know what to do with. I don't know if it's a crisis, but they do definitely have more data than they know what to do with The other side of that coin too is are we collecting a lot of data?

We don't need because we don't know what the value is. So we save everything we become orders like my garage and I always say, you know, the garage to me is like your metadata, you know, you have a garage and you put things in it. You don't really need you just throw things in boxes and you label those boxes and you put those boxes away and then you know, you go to move and that's like a new company has a new Cloud then we're gonna migrate to the clouds.

You have a brand new empty garage, right and they instead of looking through those boxes. Maybe that box says trophies on it and okay. Well, I don't want to bring my trophies.

I'm gonna yeah, I'm gonna throw that out. Well, you gotta look in that box first. It may not have trophies and they have tools they have ornaments and they have something you really want to keep in there and that's the profile and the meditated and make sure that we move what we really want and what we're really using if there's children's clothes and then you don't really need anymore.

Don't move those to the new place right? Keep those guys. Now, you don't have a brand new shiny garages the same junkyard before so I think a lot Organizations when they do their migration, it really should be a catalog first approach to make sure that they're bringing things that are that are valid but it's good good data still it's working.

It's being used and it doesn't need to be kept for compliance. All right. Hey guys, you heard it here first first, you gotta find the data and then you got to figure out which of that data is actually quality enough.

They're worth refining and then you got to actually refine it and then you get you know, the opportunity to maybe start cooking with some gas but short of that. Well figure out how we go along as always. Hey Steve.

Thanks for being on the show. My pleasure. Appreciate.

Have a great day. All right back to you guys in the studio.