The benefits have been mainly from the faster adoption or the ease with which I have deployed this for consumers or my engineering teams. The pace with which I can bring up an infra, that is, the time to market, has been important for me. So that is fast. Again, it's more time-based only, but, yes, the ease with which I was able to migrate from on-prem to Google Compute Engine was quite seamless without having to worry about whether my workloads would continue to run the same or not. So, that was a seamless experience. 

Another one I would specifically highlight is the fact that, from a perspective, most of the cloud providers give that, but I probably am a bit biased towards Google and the way in which it has its storage layer linked.  So, the persistent disk for the local SSDs provides better throughput as compared to some of the other providers, plus it now has these GPUs and TPUs as well.  So that helps a lot.

The main motive for choosing Google Compute Engine is pricing. 

Google Cloud Datalab has made it really easy for me to perform data exploration and analysis. Previously, I would have to wait for my data to transfer to a local server with a SQL database before beginning exploration. With larger data sets, the entire data transfer and analysis process could take hours if not days. With Google Cloud Datalab and Google Cloud Platform, I can receive processed results in seconds, not hours or days, without significant delays transferring data, since my data is already in Google Cloud Storage and easily accessible for consumption in Google Cloud Platform. In addition, I am starting to leverage the Google Cloud Machine Learning APIs built into Google Cloud Datalab to receive better insights on my data.

It has dramatically increased our costs over running dedicated hardware.

The GCE, with its versatility, allows us to create virtual machines for any situation. We can either put an application in production, or run software like Sonar to guarantee the quality of the code.