Reltio Connect

Chris Detzel (00:05):

Welcome to another Reltio Community Show. My name is Chris Detzel and today's topic is Reltio Integration Hub with Sajid at Advarra. So Sajid, he's the Senior Director of Data Science Platform at Advarra. By the way, this is going to be a fireside chat, so should be really fun. Ayan Basu, he's here again talking to Sajid specifically around Reltio Integration Hub. He's our senior director of product here at Reltio and I am the Director of Customer Community and Engagement.

(00:43):

So as usual, please keep yourself on mute. All questions should be asked in the chat or feel free to take yourself off of mute and ask. As usual, I will ask those questions that are put into the chat and then we can go from there. So we will be recording this and posting this to the Reltio community. So today's call is the 28th so we have Sajid and Ayan so really excited about that on Reltio Integration Hub. On the 25th of August, so we're going to skip a few weeks of shows just because we're in the summer and we'll get it started again on the 25th of August. This one is another Reltio Integration Hub to automate Reltio workflows. Should be a lot of fun on that.

(01:37):

And then, on the 29th of September, we have blazing trails and master data. So stories from women succeeding in master data management. I do foresee September being a busy month with shows. I just haven't had a chance to schedule those as of yet. So Ayan, I'm going to stop sharing. I'll let you manage this piece but Sajid will say a few things here shortly but I'll let you share your screen and get us started.

Ayan Basu (02:12):

Awesome. Yeah. Thank you, Chris. Hi, everybody. Thank you for joining and spending some time with us. So we have our special guest here today, Sajid Sayed from Advarra. Today's session will be a really interesting one because Sajid has graciously agreed to talk to us about his experience using the Reltio Integration Hub. Some of you might be aware of what the product is. It's our low code, no code platform designed to dramatically simplify and accelerate the process of building integrations and connecting Reltio Tier Enterprise Applications. And so, Sajid will talk about that and we'll have a fireside chat. I'll ask him a bunch of questions and also, you'll have an opportunity to ask questions as well. So without further ado, I'd love to jump into this but I would love for Sajid, if you want to do a quick introduction about yourself, Advarra, the company, and I'll share a couple slides that can showcase how you guys use Reltio and then more specifically Reltio Integration Hub. So Sajid, over to you.

Sajid Sayed (03:19):

Thank you so much, Ayan, for that introduction. So my name Sajid Sayed, I'm a senior director for Enterprise Data Science Platform at Advarra. Advarra is in a business on making clinical trials safer, smarter, and faster. We have two lines of business. We have a consulting business where we have a independent review boards to manage the clinical trials. And then, there is also clinical trial management software that we build, that's a technology side of the house. Currently, I am in the data science space and I've been with Advarra about two years now and I've been tasked to build this data science platform for Advarra.

Ayan Basu (03:57):

Okay. Awesome. So I'm going to share my screen and if you could talk just a little bit more about your data platform vision and then we'll jump into Reltio Integration Hub.

Sajid Sayed (04:09):

So the vision for the data science platform is to turn data into assets. To do that, we were asked to build this data platform with the objective of creating reporting analytics, help create standalone products and data APIs that we could service out to our customers. To do that, we established these four steps to help us get there. First, we wanted to create a data right and governance body, right? For that, we wanted data stewards, we wanted to create data governance council. Now, Advarra has growth through acquisitions. So what we also wanted to do was unify Advarra data products, right? Curate the data, profile it, cleanse it, master it, and then publish it.

Ayan Basu (04:56):

Okay. Perfect. And then, on this slide it would be really great to just understand how Reltio Integration Hub fits into this overall diagram.

Sajid Sayed (05:07):

So Advarra, like I said, has done a lot of growth through acquisitions, right? Some of the sources of its transactional system, some are multi-tenant, some are single tenant transactional databases. Our objective was to ingest the data, bring all that data into a raw zone. This is where all the governance was coming in and this is where we were cleansing the data, profiling the data, and after this, we do transformation within the Snowflake earlier. And then, through Reltio Integration Hub we move it out to Reltio.

Ayan Basu (05:42):

Okay. Perfect. So we're going to come back to this diagram in a little bit and dive into a lot more detail. Obviously, the audience here will be able to ask some questions. So thank you for that overview of yourself, Advarra, and what you're trying to accomplish. So I'd love to ask you, how did you first become familiar with Reltio Integration Hub at the beginning of this journey?

Sajid Sayed (06:08):

All right. So I think the architecture diagram that you saw, right? That's how we started. We knew our sources of our data. We had to bring that data to our data cloud for which we had choose Snowflake as a platform and we had an ETL tool that was ELT2 that was bringing the data to this particular raw zone. Now, we had to do the last hop where we had to move the data to Reltio, right? Unfortunately, the ETL tools that we picked, and most of the ETL tools, were moving the data from the transaction system to the warehouse but not from the warehouse to the transactional system, right? So that was the problem that we wanted to solve for and any integration tool that we are looking in the market would cost us about $100,000 to do that and we need super specialist skills to staff for that. So we reached out to Reltio with our problem statement and Reltio introduced us to RIH over here which was native, which came out of the box with our Reltio implementation, and that's how we were introduced to RIH.

Ayan Basu (07:14):

Okay. Perfect. It looks like Reltio Integration Hub was the solution that appealed to you early on. What was it about the tool itself that you felt like at that point in time would help you solve this problem? Was there specific aspects of it that really stood out to you?

Sajid Sayed (07:36):

Yeah. I think a lot of it. First, it is native to Reltio. So our source of data was Snowflake, our destination was Reltio, what we are trying to build at that point of time, right? We are still in our early phase of our implementation and we are part of a data science team so we didn't want to purchase this integration solution to just solve for this niche problem that we have. We were able to get the data from all different sources. So RIH came native to Reltio so there was not a whole lot for us to do to enable that. It was a local environment, so again, not a steep learning curve. It came with a pre-built connectors for Snowflake and Reltio so that made our job easier. So I think those were our conditions, those were the earlier insights that really helped us make the decision.

Ayan Basu (08:33):

Okay. Perfect. And so, let's say Reltio Integration Hub was not available to you. You said it would've cost you about $100,000 for an alternative solution. What would that alternative solution have looked like and how would you have gone down that route?

Sajid Sayed (08:52):

Yeah. So we were looking at a lot of integration platforms to help us do that, right? The other problem with some of the integration platform were that we would have to own and manage those systems. Now, we got Snowflake as a SaaS-based solution whereas we got Reltio which is SaaS-based. Now, just for this problem, small problem statement, having an integration platform and owning and managing maintaining and the cost for it would just not work for us and we are trying to run away from that. We were not in an integration space. We were not creating a center of excellence around data integration. We are a data science team and we wanted to stay in our zone.

Ayan Basu (09:33):

Got it. No, that's perfect. And then, as you think about Reltio Integration Hub and the points that you mentioned, the fact that it's low code, not a steep learning curve, what did it look like initially for your teams to get used to the tool? How did they learn what to do early on in the journey? Can you talk about what that initial experience was like for the development teams?

Sajid Sayed (09:56):

Sure. So I think thanks to our CSM team. First, they introduce us to the platform, right? They gave us an overview of the platform. Once we knew that, then they referred us to a documentation for Reltio. So we heavily relied on that documentation to upscale ourselves. Again, it's a local platform so it didn't took us a lot to figure things out. Just from a navigation perspective, we just needed to understand where is what. And then, whenever we were trying to build this, if we would run into a trouble, we would either log a support or reach out to a CSM team. So it was fairly straightforward for us to do the implementation.

Ayan Basu (10:35):

Okay, awesome. And then, what did the implementation process actually look like and more specifically, would you be able to give the audience a comparison of coding up that entire solution, writing hundreds, thousands of lines of code versus using the Reltio Integration Hub and putting a solution like that together?

Sajid Sayed (10:57):

Yeah. So I think just from an implementation perspective, first thing what we had to do was make sure Reltio Integration Hub was enabled in our environment, their test production. Once that was there, we had to make sure our connection endpoints, right? So I think whatever work that we had to do was to ensure Reltio connection came out of the box, we had to configure for Snowflake connection. Then I think one thing for us to be aware of was best practices so RIH will have its own best practices. We had to understand from an organization widely what were our internal best practices that we wanted to implement. And then DevOps perspective, how do we make sure that this integration moves between our dev test and production environment?

(11:46):

So these were some things that we had to be mindful about. Again, from a coding perspective, there was hardly any but what we still had to learn is what does looping mean, what does logging mean, and how to go about that. When we are handling a large amount of JSON files, how do we break it up? So there are things that you need to know, not everything will happen magically but then we didn't have to code around all those.

Ayan Basu (12:15):

And then generally, after the entire solution was built, what was the overall feedback and experience, especially after the team learned the tool, learned the inner workings of it, what was the general feedback that you came across?

Sajid Sayed (12:34):

So with everything else that we did, right? So I think this was the last piece of the puzzle that we were supposed to complete and we completed this four times faster than we had actually anticipated that we will reach a problem. So from an experience perspective, it was much easier for us to do. And then, what we did not have to do is we didn't have to go and hire a super integration specialist. So we had a data engineer who had other jobs and was quickly able to come, get upskilled, do this integration, and go back to the job. So for us, just from going to market perspective and keeping the system up and ready, there was hardly any maintenance or production support that we have to set up around this.

Ayan Basu (13:24):

That's great. You said it was about four times faster. What are the key things that accelerated that? Obviously, the coding aspect is one. What are the other things that contributed to the speed, do you think?

Sajid Sayed (13:38):

Everything is UI based, right? So everything is UI based, it's not a separate entity to Reltio, right? What we had to do is we have to understand the Reltio data model. We had to understand the destination data model but because it came in connected with Reltio, I think that part was very easy for us. There was no environment management. If we had this different environment, we had to make sure we are managing all these different environments which we didn't have to do so we have zero production support team for integration. All we do is logging. So as part of that RIH piece, we log. So we need to understand our system from the entire data pipeline perspective so what we do is as part of that RIH process, we log everything that's happening into the system and now, we have this central portal where we see this entire data transformation that's happening. And so, it just makes it easy. That's the only focus that we have. We don't really have to worry about the platform.

Ayan Basu (14:39):

Okay. No, that's great. I see some questions coming into chat which is great. So I do want to read out this first question. So how is the integration hub performing in terms of latency, of data loads, and any challenges and performance that you ran into? Can you talk about the performance testing and data load aspect of which you encountered?

Sajid Sayed (15:03):

So I don't know if I have the benchmarks that we did but I think what we did as part of our first implementation, everything was in batch mode. I think what we can do if we had to do this again and as we improve the process, now we can actually create real-time integrations with RIH, right?

Ayan Basu (15:25):

Yeah.

Sajid Sayed (15:25):

So that is a performance that we have to be mindful about. But if you start doing this real-time, and like I said, we are getting the data from the source to this Snowflake area, we are doing a lot of transformation, there are certain data quality checks that we can do actually in flight so I think that's where we are leading into. Our future states will actually really have that.

Ayan Basu (15:51):

Even though you did batch mode and batch loading and you talked about going real-time which is great, is there anything you encountered that would create any pause or concern in terms of performance or did you feel like the platform itself met your performance criteria?

Sajid Sayed (16:08):

Yeah. So it did meet the criteria but I think just something to be careful about, something that we are still in the learning phase, right? Right now, when we are extracting data out of Reltio, you get this major JSON file. So what we'll have to understand is how do we loop through that? How do we compartmentalize that and don't take one big file but a certain amount of data at one time, pass through it, and move it? So there are certain things that we learn as part of the implementation, it's not performance drawback more, it's something for us to learn and know how to use the platform.

Ayan Basu (16:43):

Yeah, yeah. That's great. So another question we have in the chat is how much time did it take from the decision of using Reltio Integration Hub all the way to going live into production, what was that timeframe?

Sajid Sayed (17:03):

So just from the Reltio Integration Hub, it was four weeks worth of effort.

Ayan Basu (17:09):

Four weeks worth to build it, right? And then, test production. Yeah.

Sajid Sayed (17:13):

Correct. But everything else was not just linear because there were other program and project decisions that we had to make, but just from a dev effort, it took us four weeks to do that. Just a disclaimer, we don't have any complex transformation in that. It's a simple, straightforward pipeline because we had done major of the transformation within the Snowflake area and this was just the last push that we were doing. As part of that last push, we discovered, hey, there is a lot that we can actually do in this platform.

Ayan Basu (17:44):

Yeah, absolutely. If you had to code the solution, obviously it would take much longer than four weeks. Is that an accurate statement?

Sajid Sayed (17:54):

Oh yes, definitely. The other problem is then that knowledge is with that particular developer alone. So now, if we run into any issues with Reltio A, it's very easy for somebody to look at it and figure out what's happening. But then we can also go to RIH support forums and customer support and figure out if there are certain things. But we didn't want to get locked into whether IP is just with that particular developer and we cannot use other people in our data engineering team to support the solution.

Ayan Basu (18:25):

Yeah, absolutely. You did mention the fact that supporting the solution long-term maintainability, that's much easier with integration hub as opposed to separate infrastructure, separate integration, that is a whole IT development team that's dedicated to supporting that.

Sajid Sayed (18:45):

Correct. From a data science team, we wanted to go to market really, really fast. As many, if you start having these different technologies, we need to have super specialists to support each of them and then DevOps process in each of them. If there is something with a larger memory, we need to increase that memory. And then, if there's a difference data source, we need to create tunnels or whatever we need to get to that data source. So now, we have avoided all that by going to a SaaS platform which is natively integrated with these sources and comes with pre-built connectors for most of them.

Ayan Basu (19:19):

Yeah. So another question, is there any training on RIH provided by Reltio, videos, documentation? We do have an extensive list of links, documentation, videos, webinars like this where we talk through Reltio Integration Hub. So if you go to the Reltio YouTube channel, you'll see a lot. There's Reltio academy, there's links to documentation on our website. But then also, question for you Sajid is how did the training happen on your team? You mentioned going to documentation, did they also go through Reltio Academy, the Workato Academy training? Can you talk a little bit about how they trained themselves on this tool?

Sajid Sayed (20:01):

I think so, yeah. So I think the same material that you spoke about, I think the biggest help that we had was from the CSM team, right? So you guys gave us the jump start. So once we were over there then it was the help documents were actually good enough for us to solve most of the problems. Like the large payload issues that I was talking about, those are something that we actually figured out while doing what we are trying to do. For those issues, we love the support and the support team helped us out with solutions for that.

Ayan Basu (20:33):

Got it. Next question is in the architecture diagram... Let me pull that one up. Give me one second. It looks like there's both cleansing and... Actually, give me one second. I'm just going to pull this up before I continue on with the question. So yeah, there's a separate cleansing that you do here but then the question is, in Reltio you also have cleansing so can you talk a little bit about this versus what's done in Reltio itself?

Sajid Sayed (21:11):

Right. Mind you, all of these things, we did in separate phases. So what we did was, in this particular case, when we got this data from these different sources as the first part of that activity, we did data profiling and we did a lot of data cleansing over year and transformation so that at that point of time we had not actually decided our MDM solution, right? So when we started this whole implementation, we were thinking about, "Hey, can we just do this natively ourselves?" We have this data over year, we cleanse it, and then we very quickly figured out that we need a MDM solution which has a data model in the industry that we need to suffice it. So this is what we started with and as we are maturing a model, like I said, our goal is to get Reltio closer to the data sources and drop the in-between applications.

Ayan Basu (22:10):

Perfect. Thank you. All right. So going down a list of questions and this is getting into a level of detail which you might not have awareness to, but we talk a lot about those concept of tasks, right? The steps in between and all of our customers get a default number of tasks for the recipes. Did you have any thoughts around optimization of tasks in the recipe? How your team viewed that? Anything you want to share around the compute resources used?

Sajid Sayed (22:47):

Yeah. So at this point of time, we had some discussion around that but for whatever we are doing in our space, we were hardly any of those limits. So we are not actually in that space where we have to do any efficiency games in what we are doing right now. So we actually didn't dive any deep in that part so far.

Ayan Basu (23:07):

Yeah. Okay. Perfect. Another question is and a lot of people do ask this Reltio Integration Hub, is it for batch? Is it for realtime? From what you just mentioned Sajid and even things that we have talked about from a Reltio perspective, you can use Reltio Integration Hub both in a batch and real-time manner. It offers you that level of flexibility. I don't... Have you gone into any of the real-time integrations just yet or is it too early to provide any thoughts there?

Sajid Sayed (23:40):

It's too early but it is... Like from the call out that somebody had, right? There's just so much cleansing that you are doing in between, right? Now, there is so much shift that we can do in throttle while creating a field, check this, check that, make sure that if there is a golden record already existing, use that instead of creating one and we can do that all real-time. So I think that's the world that we have to be in. But as we are, like I said, we have a lot of growth due to acquisition, we have to also figure out how do we create pipelines into all of the system and we really think RIH can play actually a big role in that space.

Ayan Basu (24:18):

Perfect. Actually, that's a good segue into the question that I meant to ask you earlier. Can you provide some insight into what the longer term plan looks like with RIH? Obviously, you've had good experiences so far. As you think about one year down the road, two years down the road, can it actually be that central connectivity integration platform into your larger, broader ecosystem efforts?

Sajid Sayed (24:47):

Yeah. No, I definitely think that's the area that we need to invest in, right? Bring Reltio close to source as much as possible, move most of it, whatever we can, from batch to real-time and to connect all these different transactional applications that, like we said, we have inherited through growth, all right? So I think that is the biggest business value that we can get.

Ayan Basu (25:15):

Awesome. There are some questions coming in chat. Does it integrate to Salesforce, Veeva CRM? Yes, it can integrate to those applications. We do have a specific connector for Salesforce that Reltio provides which I would recommend looking at first and then Reltio Integration Hub can be used for additional use cases on top of that, that the Salesforce connector cannot solve today because it's purpose built for certain use cases. But that's really the beauty of the Reltio Integration Hub is this vast amount of flexibility it provides to integrate to all these different applications.

(25:53):

Sajid, you mentioned one of the biggest appealing factors was the fact that the platform does come with a lot of pre-built out of the box connectors that you've already used, the Snowflake connector that's in there and there's others that you can look at. As you looked at these other platforms versus Reltio Integration Hub, what did that look like, the pre-built connectors out of the box? Was that a criteria for comparison as you looked at different systems?

Sajid Sayed (26:23):

No, definitely. So we have a very heterogeneous environment. We have Oracle SQL server, postgres, so we have all of that. So for us to make sure that there are connectors for... And then, different versions of Oracle, right? We have some of their pluggable database in Oracle so we wanted to make sure that we have connectors for all these environments so I think that's what very critical for us. We didn't want to create a single pipeline for one platform for each transaction system that we had.

Ayan Basu (26:52):

Yeah, absolutely. Yeah. Just one stop shop for everything would be obviously the ideal way to go. So looking back on your integration hub journey and experience, what would you have done differently? What were the learnings? What would you recommend to the people that have joined this call that have maybe haven't used Reltio Integration Hub? Any advice or guidance you would have for them?

Sajid Sayed (27:16):

So I think we looked at RIH very late in our journey and maybe it was just released about that time too, right? So we just started with our problem to solve. So we say, "Okay, I just want to connect from this source to this source," and that's about it, right? But I think for anybody starting fresh, I would like to tell them that, take a step back, look at what RIH has to offer from a whole platform perspective. There are so many transformations that you could do in flight so see if you can use that, right? There are so many jobs that you can actually schedule in there versus... Some of the challenges that we have like Snowflake is a great platform but when you start putting tasks and all that in there, it's just little hard from a UI perspective to know which bad job is running where. So there's a lot of those that you can shift, that RIH can do that for you.

Ayan Basu (28:07):

Okay. Yeah, that's great. Great tips. We do have another question. Do we need an additional license for using the connector application like SAP, ECC, or HANA? So that's one of the benefits of this platform is you don't need additional licenses. Any connector that's on the integration hub, you can just use. It's all available to you at no additional cost. For setting up to on-prem environments, there is an agent that's on the platform that would need to be enabled for you to connect to on-prem applications but it is possible through integration hub. Anything on the on-prem side that you looked at connecting, Sajid, as part of your RIH efforts?

Sajid Sayed (28:51):

I think we will have to. At this point, not but some of our Oracle databases are on-prem so I think we will have to address that when we try to build our real-time integrations.

Ayan Basu (29:04):

Yup. Yeah. We do have other customers that are actively integrating to on-prem applications today using integration hub. So another question is, last time we had to use a custom interface to get data from CRM to Salesforce... There will be no issues using the RIH integration for Salesforce with merging or any other integrations. We do have customers that are doing that today but if you have specific questions around that, we can have a discussion on it offline. But so far, we haven't encountered any blockers or obstacles in terms of integration hub being used to solve all these various use cases on these platforms.

(29:54):

Another question is, is RIH used internally? By the way, RIH is the acronym for Reltio Integration Hub, probably you already picked that up. Is it used for Snowflake integration? So the Snowflake connector that we have does not use Reltio Integration Hub. It is purpose built. There's a separate application that we developed. It's a separate data pipeline hub that we created and those connectors are specific to analytics platform and it does not use Reltio Integration Hub as the underlying infrastructure. However, we do have a specific connector into Bureau van Dijk that does leverage the Reltio Integration Hub and it's a series of recipes that have been built to enrich your data within Reltio using RIH. And then going forward, there's the capability to build connectors using SDKs on the platform itself and other tools and capabilities on RIH. A question just came up in my mind, Sajid, is have you and the team looked at building any custom connectors on the RIH platform using DASD case?

Sajid Sayed (31:08):

Not at this point. Yeah. Not at this point.

Ayan Basu (31:11):

Okay. Okay. Yeah. That capability is there which opens up the flexibility even more. So that was the initial set of questions I had. I covered all the ones in chat. Any other questions from anyone? So I think we covered a good amount of topics on the RIH and specific to your use case. Any last words that you have, Sajid? Anything on your mind that you want to share that we haven't covered?

Sajid Sayed (31:50):

No. I think for us, especially from a Reltio perspective, we are thinking about integration, we don't have to go out and buy an integration platform. Even if you have an internal integration platform, you don't have to rely on that team to build this integration. You don't need a super specialist data integration architect to help you with the integration. So think about these things, these are available to you to build integration. It will just make your life easier and help you get to market faster.

Ayan Basu (32:21):

That's great. Yeah. No, thank you very much. I really appreciate the input and the insight. So yeah, as Sajid mentioned, saved $100,000, four times as fast, you don't need specialists to go do this work, you can leverage your existing teams build on RIH. So I think, hopefully, all of you got to got a chance to hear it from Sajid himself and his team that went through the entire process. RIH is available for... It is available at no extra cost, there's a question here if it's a free trial. We can get it enabled. There is some paperwork that you have to sign but we will reach out to you and get that going.

(33:05):

But then, a couple other things to add is, thanks to Sajid and Advarra, we are doing a case study. We're in the last stages of getting it approved and finalized and getting Sajid to review it. This will be a published case study on our website. We will share it with you. In that way, you'll get to see in writing everything that we talked about and hopefully give you some ideas and thoughts around how you might want to build use cases on Reltio Integration Hub. As you can see, a lot of value, a lot of flexibility, and hopefully you're able to use it to solve some of your own challenging business problems that you're encountering. So I believe that's pretty much it for me and Sajid. So Chris, do you want to close this one out for us?

Chris Detzel (33:56):

Ayan, thanks so much for providing some more information on Reltio Integration Hub. It's always really good to hear from our customers that are doing the work and those things so thank you everyone for coming. Hopefully, this was helpful. I know it was for me. It was a great story. Sajid, thanks so much for coming and answering some of these really tough questions that Ayan put together. No, but it was really good so thank you everyone. I'll stay on for a little bit, would love for a little bit of feedback in the comment section and I'd like to share that with the folks here internally. So until next month, our next show will be on, I think, the 25th. So thank you everyone for coming. Sajid and Ayan, thanks so much for hosting the show today.

Sajid Sayed (34:44):

Thank you for having me.

Chris Detzel (34:45):

Of course.

Ayan Basu (34:46):

Thank you, Sajid. Thanks, Chris. Thanks, everybody.

Chris Detzel (34:48):

Thanks, everyone.