Thanks AJ! Traditional data warehouse projects use many software products--no doubt you've used some of these [enumerate: ER modeling tools, metadata repositories, data profiling tools, MDM components, workflow tools, and probably the largest component, ETL tools]. Today I'm going to talk about how you can eliminate some or all of these products from your environment in a way that allows you to deliver in a much more agile manner.
How can we do this? Kalido's core solution is the Kalido Information Engine which sits in the data warehouse automation space. Using a high level business model, Kalido converts that model to metadata that drives software automation infused with industry best practice to create the solution. [highlight a few items, all the things you'd typically do in a best practice warehousing implementation, we automate]
The first product we think can be successfully eliminated is ER modeling tools. These are tools like ERwin, ER Studio, Rational Rose. How is this possible? Kalido has its own modeling component we call the Business Information Modeler. The model you create with this is a high level conceptual model, but most importantly, something that can be easily understood by the business. What we are looking at here is a small sample model,just so you can read it, but it gives you an idea. [the blue cubes are master data, so you can see customer and how it is segmented. You can see business rules like Region being optional. You can see activities or transactions linked to that master data.] The most important bit here is that this model becomes the common language between the business and IT, and many of our customers in fact print out these models and hang them on the wall of their office or cube! I'd estimate that 99% of Kalido customers no longer manage their warehouses in an ER modeling tool and they are much better off for it!
Why do I say that and Why do we take this approach? Here is a brief description of the traditional modeling disconnect
Kalido - just go back to the model, add your new boxes, press deploy, and Kalido physically makes the changes in the warehouse in a way that supports standard Dev, Test, and Production migration. By the way, this BIM modeling tool is a free download from our site. Some consultants use this to collect requirements even without using Kalido. What we charge for is the button to deploy it. :-) By automating the process from conceptual to physical (but keeping the linkage back to the original requirements), you can implement faster and be more agile for future changes. So that's modeling tools. What's next? How about metadata repositories?
Metadata Repostories. Maybe 5-10% of Kalido customers (usually larger customers) use us with some form of metadata repository (these are tools like like ASG Rochade or IBM Business Glossary). And certainly if you want to push Kalido metadata to one of these products, you can easily do it because all Kalido metadata sits in the relational database and we include views that make it very easy to get. [go through points]So we're saying you don't need a metadata repository, but don't Kalido customers need to access this this metadata? Sure they do, but they don't need to do it with a separate tool.
Now let's talk about ETL. Most experienced DW practitioners will tell you that a typical data warehouse project takes at least a year and that 80% of that project is likely to be on the integration side. So anything you can do to reduce that effort will have a massive affect in how fast you can bring in that project.
Kalido customers tell us we do just that. But how? Kalido uses more of an E-L-T approach. Rather than rely on an ETL application server to move the data, transform it, and then push it into the database server, we land the data to the database server as fast as we can and then use set based operations at the database level to handle the transformations. This is usually much faster because the database server, especially if you are running a data warehouse appliance like Teradata or Exadata, is often the biggest server you have in your company. What's more, since our data integration is tied back to the business model, a change you make in the business model should be able to be cascaded through to the operational component and allow us to update most if not all of the process automatically. For example, if you add an attribute to customer, we automatically change the staging table to add that column, and we update the load job to add the new attribute as an available mapping. It is up to you to ensure that new attribute lands in the staging table, but at that point, Kalido takes it the rest of the way all the way to the reporting schema.
In summary for ETL, this is why I'd say half our new customers since the release of Kalido V9 a few years ago haven't used a third-party ETL tool. Those that do, often find that they can get away with lower tier tool, or leverage a tool they already own (e.g. SSIS which comes with SQL Server), or specialist tools that help on the extraction side (for instance, Attivio to grab mainframe data / COBOL copybooks, IBI that has connectors to everything, etc.). This isn't to say that if you have a large investment in DataStage or Informatica you can't keep using it. You can, but the way you use it may change. These tools can be used to populate Kalido managed staging tables (the E), or orchestrate Kalido jobs using their scheduler, but the goal is to eliminate the ETL hardcoding to reporting schema tables (which change frequently and require ETL updates) and again, to leverage the speed and power of your underlying database platform.In a case recently, a customer was able to save a 7 figure annual ETL maintaince bill by switching to Kalido. So if you can reduce your ETL footprint, you can not only improve performance, but find some real dollar business cases in there.
The last area I want to cover is Master Data Management. We could fill a webinar with just this topic, but briefly, traditionally Garter breaks down MDM to two areas, CDI or Customer focused tools, and PIM or Product focused tools. Additionally some of the tools in this space are extremely expensive and whatever you pay will generally cost twice that amount to implement. By contrast, Kalido Information Engine _includes_ an MDM product that instead of being hardcoded to one space or another, is an every domain solution that can be used for any master data need. [AB InBev uses us for over 250 classes of master data -- in one instance -- and we have other customers mastering a similar amount.] Customer example:Daymon – More than 60 different master data types: Customer, Supplier, Employee, “Item”, Brand, Location, …AB-InBev – More than 250 domains (classes of data)Smith & Associates – Both customer and product
Also, our MDM product is driven by the same BIM model we saw previously. Model it, publish it, and you have built the system!Smith & Associates – Many to many relationships Managing product hierarchies - part supplier kind of thing - multiple relationships around parts - parts around manufacturers - they're acting as a distribution source. Trick is that the same part Daewoo might be used by both HP and Dell and have different part numbers. Likewise, Dell might have a secondary supplier with Samsung which has the same part number.
Further, you don't need to purchase a third-party workflow component like Lombardi to make it work. Kalido includes an MDM specific workflow component so you can tie in non-technical data stewards into the data remediation and approval process--with no coding. [You can start out small with just simple workflows, perhaps a state that captures invalid data for correction, and another for approval to larger workflows like this media company that routes data to 14 different departments.]Time permitting: Data Profiling - technical profiling replaced by business oriented reverse profiling, you quickly define what you think the model is for the components you need, perhaps speaking with a source system expert, and immediately load a subset. Kalido highlights errors based on your assumption and allow non-technical users to see the values and suggest changes. These changes could be model changes, or rule changes, or both.Virgin Media – sophisticated – 14 different departments that need to be involved in lifecycleUS Bank - Workflow managed financial plan and budget forecast data, replacing current spreadsheet managed data, via MDM
In summary, we've highlighted at least four classes of software today that we think you can eliminate from your stack by replacing it with another in the data warehouse automation space, specifically the Kalido Information Engine. With that, I'd like to hand it back to AJ for the final survey question and any questions that have come up in the meantime. AJ?
Thanks for tuning in today. If you would like to learn more about how you can reduce the cost of your iterations while generated improved business value from them, you can start by taking Kalido’s on-line business agility assessment.If you are planning on attending the TDWI BI Summit this August, you should register for the sessions from Ralph Hughes that explain the agile tools behind these shorter release cycles.Finally you can see this in action by calling or e-mailing Kalido to request a demonstration.Thanks and I’ll turn it back over to Ben.