Just a quick architecture diagram here to kind of get a lot of these terms cleared up. So, starting with the left. User will query BigQuery. A unit of work in BigQuery itself is called a job. We'll revisit the job when we talk about BigQuery pricing later on. Jobs run on a very fast analytics engine that was developed internally at Google and then made available as a service through BigQuery. And then those query jobs are then mapped to the underlying data, which is fully managed behind the scenes in those tables. And then, walking back the other way, all the way at the bottom there, you can ingest data into something like Google Cloud Storage if you wanted to. Or directly into BigQuery if you wanted to, and then have that be available for analysis. But the key takeaway from this slide is at the top, you have the BigQuery Analytics engine in that one box, and then you also have the BigQuery Managed Storage. So you have this powerful query engine, and you also have this replicated scalable storage for all your data that is being stored. So it's actually two technologies, or two services in one. So, Google BigQuery is that managed storage piece, which is scalable and it's the same technology that stores a lot of Google's product data, right? Think ads, Google email service, Gmail. But it's also that really lightning fast analytics engine, SQL engine, and it's built on the massive evolution of Google Technologies over time. The relentless march, if you will, to keep performing better and better, because Google is naturally incentivized because of the amount, massive amounts of data that it has. So let's talk a little bit about that relentless march. So Google loves to innovate data technologies. So there's a lot that are focused on here. One of the words that may immediately look familiar to those who have been around the big data block for a while is MapReduce. In 2004, Google Research actually came out with the white paper that became MapReduce, and then open sourced it, which was then used as the foundation for Hadoop, which is that massive parallel-processing, right? Bits of data mapped with tasks and then processing all of that in parallel. Not content with that, in 2008, we released the Dremel white paper, which is processing queries over smaller chunks of data, but doing it massively in parallel, and having that done through sequel. And that, the Dremel Technology, plus Colossus, which is that massive hard drive in the Cloud, those two technologies form the basis of what was then BigQuery and Google Cloud Storage as well. So as you can see, Google has opened up those technologies to you as part of the Google Cloud platform, and continues to innovate. If the other technologies here interest you, Dataflow, again, is one of those data engineering tools where you can build those massive data pipelines, ingest streaming data, and batch data and then dump it into BigQuery. If machine learning is your game, learning things like TensorFlow as part of additional courses, is also one of those great technologies that's available through Google Cloud Platform as well. And if your ultimate end result is to get to machine learning, stick around for the third course in this specialization, where we'll cover a lot of the initial introductions to some of the tools.
