Joe Caserta: Comparing Cloud Offerings and Understanding AI
In this podcast, Wayne Eckerson and Joe Caserta discuss data migration, compare cloud offerings from Amazon, Google, and Microsoft, and define and explain artificial intelligence.
Joe Caserta is President of a New York City-based consulting firm he founded in 2001 and a longtime data guy. In 2004, Joe teamed up with data warehousing legend, Ralph Kimball to write to write the book The Data Warehouse ETL Toolkit. Today he’s now one of the leading authorities on big data implementations. This makes Joe one of the few individuals with in-the-trenches experience on both sides of the data divide, traditional data warehousing on relational databases and big data implementations on Hadoop and the cloud. His perspectives are always insightful.
Key Takeaways
- 100% of Caserta’s clients are moving to the cloud
- Google and AWS are more similar than Microsoft
- Google adoption is growing, and they’re investing in an enterprise sales force
- Serverless is big because it removes all infrastructure problems from the customer
- Machine learning is not artificial intelligence
- Artificial intelligence is defined by it’s ability to answer the second question
The following is an abridged, edited transcript of the podcast
Wayne Eckerson: Is it possible to migrate 100% of your ecosystem to the cloud?
Joe Caserta: I think it’s more than possible. I think it’s probable, and if I had a crystal ball it would probably tell me that a large percentage of every organization will eventually move to the cloud. We just helped a very large publishing company close down their data center. It was a big event actually going and hitting the power button in the racks and actually turning off all the servers they’ve maintained for the past few decades. The days of physical servers and racks and maintaining your own infrastructure is a thing of the past, at least in my world. I’d say five years ago we were 100% on-prem solutions. Three or four years ago we were half in the cloud, but starting last year maybe 80% of our projects were. This year 100% of our projects are either migrating to the cloud or building something from scratch on the cloud.
Wayne Eckerson: So in my world I’m seeing a lot of Azure implementations. It seems like AWS came in as infrastructure as a service and now it moved up the stack, but Microsoft came in as platform as a service, so a more packaged offering. I don’t know much about Google Cloud Platform, so maybe you could differentiate those three platforms.
Joe Caserta: I think Google and AWS are more similar than Microsoft. If you are a Microsoft shop using SQL Server and all the ancillary tools around SQL Server, then upscaling to Azure makes a whole lot of sense. If you’re more enterprise, coming from the Oracle world or a little bit more advanced technically where you were doing Hadoop already, then those clients are typically going to AWS or Google.
Wayne Eckerson: So what makes Google a real hot property? Why are people turning to them? I haven’t seen them at all and they’re certainly a third in market share.
Joe Caserta: I think in market share they are third, but in growth, I don’t know statistically, but we are seeing tremendous adoption of Google. We’re actually seeing some clients go from AWS to Google, never the other way around.
Wayne Eckerson: So what’s driving the success of Google?
Joe Caserta: I think they finally invested in an enterprise sales force. So they are giving better customer service to their customers, rather than a swipe of the credit card and lots of luck. They are actually doing a very high touch sales engagement, which is the way enterprises tend to want to do business. So, that’s helping a lot. And just the product itself! Google Big Query is really an outstanding product; the ease of use, the performance, the maintainability, the fact that it’s serverless, which is a new paradigm that everyone wants to get involved in. It’s hands down probably the best solution out there for accessing, querying arbitrary big data.
Wayne Eckerson: Artificial intelligence is now the most hyped thing out in the market place. What is it? Everyone is using the term, but I don’t think anyone really knows what it is. What would you say artificial intelligence is?
Joe Caserta: Today, it’s still pretty ambiguous. I think it means different things to different people. Even the term data warehouse, 30 or 40 years after the term was created, people to this day still have their own definition of what a data warehouse is. I still hear arguments about whether it should have operational data, should it be aggregated data. After decades, people still don’t agree with what it is.
Since AI just started hitting the market in the past year or two, I’m not surprised it’s not a concrete defined entity and that there’s ambiguity around it. I describe it as a set of processes that directly affect a individual’s or human being’s behavior. So, if you’re trying to interact with a product and the product does not have AI, it’s going to take a little bit of interrogation in order to get the product to do what you want. If the product has embedded AI, especially if the product was built with AI from the ground up, then your interaction with the product will be much more efficient and much more pleasing to the consumer.
I should say the consumer could be anything. It could be a person on an app, it could be someone trying to buy a product, it could be someone trying to build something, it could even be an artist trying to create a painting. AI is going to affect every aspect of every individual on the planet. It’s going to be that pervasive. And not 20 years from now, I think in the next few years. It’s going to happen so fast. Essentially, every single one of our clients is at least asking about it, and some of them are actually implementing it.