I think the authors were up-front that some of the issues they encounters (for example table optimization) could have been done in Redshift. What I found intri

I think the authors were up-front that some of the issues they encounters (for example table optimization) could have been done in Redshift. What I found intriguing about the cost savings was that it was achieved on Amazon as opposed to going on-prem where you can push costs down further especially if the workload is reasonably stable. It's therefore likely that there are still substantial savings that can be squeezed out. That gap gets bigger if you scale to large cluster sizes. In thinking about costs it's also important to remember you can run a data warehouse perfectly well without a cloud management plane. It really depends on the data warehouse complexity, how many other things the team has to manage, and their comfort level with the technology. Redshift--to its credit--has so changed the conversation about data warehouse operation that we tend to forget this. Disclaimer: I work for Altinity and the article came from our blog. We did not write it and the authors are not customers.