This is the last of my Exterior design topics. Again the underlying question was how does one design for an ever changing set of user requirements. We have already looked at two of the three known attributes: Different queries against same data and increasing frequency of loading the same data. The last known is that regardless of all else, you are going to have growth and you need to be ready to deal with it.
Editor’s note: Rob Armstrong is an employee of Teradata. Teradata is a sponsor of The Smart Data Collective.
Growth is coming from many different areas; much of it is simple inertia of the existing data. If you are growing as the company then you are getting more customers, customers are using products more or buying more, or you may be opening new locations. All of this means that any data that is related to these aspects will naturally increase. Are you ready for it? At one customer I was working with, they were concerned that the batch window was being exceeded. The load was suppose to finish by 6:00 AM and it was running over. The sad part is that after just a little investigation it was discovered that the load had been increasingly longer over the past few months. This “sudden crisis” was entirely predictable.
So are you currently tracking and understanding your load times and any deviation from the expected? The next question would be whether or not your current architecture is ready for the growth. If you have a process that is scalable and is simply a matter of applying the necessary resources, then kudos. If not, then start planning for it now.
More Read
The other part of growth is harder to predict but if you are successful it will happen quickly. This is the constant request for new data sources to be included in the environment. This has a direct relationship with our first known of users wanting to ask new questions. Do you already have a roadmap for what data is coming next? Do you understand the workload (load processing as well as query capacity) that will be coming down the pike? Will this new workload be the existing users (which means some current work will be dropped) or new users (which means a whole new level of processing)?
To connect this topic with our second one of frequency, the other known is that the users will not only want the data loaded more frequently but at a more granular level. One example is from retailing. Initially data was loaded at an aggregate level of items per store per day. As retailers have gotten more sophisticated in their analytics and processes they now want individual transaction level. This is an order of magnitude difference in terms of data volume. Is your current solution ready to grow exponentially?
The interesting part is that this growth occurs because you are successful in your current efforts. The users are getting smarter and now demanding more. If you have to tell them that it will take 6 to 12 months to bring new capability to the table, all your success will be forgotten and the system will start to slowly wither as users go elsewhere.
OK, homework assignment (as I know you are all not busy enough!). Take a look at your current environment and just run a paper exercise. Pretend the users have now asked that the level of data be taken one or two steps closer to “atomic level.” Additionally, pretend the load frequency needs to be twice as frequent as you currently have implemented. Where are your roadblocks? How do you get rid of them?
This is a simple but very enlightening exercise as, once done, you can go to the business and executive levels and start preparing for the inevitable.
To end the interior / exterior design analogy. When we rebuilt our house I knew that a wall mounted TV was going to happen. I knew we would eventually put in a bar area on the deck. By understanding this we adequately plumbed and wired the necessary areas. Now we can complete the exterior (correctly and without massive rework) because we have the enabling underlying interior architecture.
I have been thinking of other topics so the blog will bounce around a bit to other areas that interest me. I am also back on the road so perhaps I’ll have to share something that pops up in my conversations with my customers and colleagues. If there are topics of interest that you have, please let me know and I would be happy to give my take on your concerns.
