“Don’t rewrite —virtualize” is the new mantra that is reforming the world of data warehousing. Let’s step back for a moment. The data warehouse is the single most important component when moving to the cloud. It’s also the hardest to replatform. Through its quasi-monopoly in this space, Teradata has become a magnet of attention.
For sure, the fact that enterprises are trying to escape 40 years of vendor lock-in when it comes to databases is nothing new. However, this time it’s different. For starters, everybody has to replatform—or risk missing out on the revolutionary benefits the cloud offers. And nobody has more than a few years’ runway to do it. Otherwise, they will be overtaken by their competition.
A naïve approach to this problem would be this: first, migrate the content of the data warehouse by moving all the data to the cloud, and then rewrite, adjust, rebuild all applications. We call this approach simply “Rewrite”. Unfortunately, rewrite is dangerous, costly, and incredibly time-consuming.
In this article we make the case to virtualize instead of rewriting. We created a handy guidebook where we discuss the mechanics and pitfalls of rewrites and contrast them with a highly cost-effective and far less risky alternative. Let’s take a look and understand what’s at stake.
Replatforming an enterprise’s Teradata footprint to the cloud is not about database technology, counter to what some might think. Enterprises need to get their data into the cloud quickly in order to survive. It is that drastic.
Consider for example retailers who want to create a personalized shopping experience. Having to build ML and AI systems on premises is cost-prohibitive Few have the resources and the requisite skill sets to create complex software platforms. Never mind the challenges around maintaining and continuously advancing that technology. On the other hand, getting the data into the cloud lets them leverage top-of-the-line ML/AI engines. A vast array of vertical-specific solutions is now at their fingertips—ready to deploy. “That’s obvious”, you think. Other verticals benefit from the cloud in analogous ways.
So, migration is not about cool technology but about the value of data. Don’t get carried away with looking at it primarily as an opportunity to clean up the toolshed. Today, migration is about closing the gap between the potential value of your data and the realizing the full potential of your data before your competitors get there first. Don’t waste time on rebuilding perfectly good applications. Set your sights on the horizon. Embrace legacy and move fast!
Some IT departments seem to have migrated from Teradata to another database or a cloud data warehouse already. But have they really? To the best of our knowledge, almost no enterprise has migrated off a Teradata Enterprise Data Warehouse at all. To get to the bottom of this seeming contradiction, we recommend this question: “Has the Teradata appliance been decommissioned yet?” In most cases, this will reveal that a number of workloads might have been off-loaded but the project stands at, say, “75% complete”. Maybe it is expected to complete “within no more than two years”?
In any case, almost complete is not the same as “done”. When going to the cloud, the appliance needs to be decommissioned. No ifs, ands, or buts. Anything short of truly complete will not work.
Teradata workloads are rich and complex. Rewriting applications in such an environment is a truly Sisyphean task. Gloomy news about aborted migration attempts and wasted resources are common.
The problem is the human psyche. Once we understand the basic technical underpinnings of a rewrite, we’d like to believe rewriting isn’t that hard. And we tell ourselves that addressing the hard problems for which we don’t have a solution just yet can be postponed to a later point in time. The fact that the vast majority of attempted migrations are late, over budget, and fail underlines just how easy it is to get tricked into this mindset.
Teradata migrations are particularly difficult. They are much harder than migrations from Netezza or other specialty appliances. Teradata workloads are larger and more varied, more applications are involved, etc.
Don’t fall prey to the “how hard can it be?” and wreck your career.
At Datometry we pioneered data warehouse virtualization. With Datometry, enterprises are now running their existing applications, written for Teradata, natively on a cloud data warehouse. Migrating off Teradata is now a real option, without risking career and enterprise. The Hyper-Q platform transforms SQL on-the-fly and synthesizes optimized SQL for the desired cloud data warehouse. Through its emulation of complex features, Datometry Hyper-Q gives applications a fully equivalent platform using the latest cloud data warehouse technology in Azure or AWS. In addition, a complete suite of client-side tools, from drivers to loaders rounds out the offering.
See the guidebook for a detailed analysis of challenges when rewriting and how Datometry overcomes these obstacles.
To learn how to transition from Teradata to any modern cloud data warehouse at a fraction of time, cost and risk, please visit www.datometry.com