I'm not sure this blog that seems to talk about the problems of NoSQL databases in general makes a lot of sense. If you read the comments, you certainly see lots of complaints, but I also think the post isn't well written. It lists the problems of RDBMSes as possibly deleting all tables while changing a key or being unable to add a column easily. While I don't know about all RDBMSes, I don't know any that could lose all data with anything less than DROP DATABASE or TRUNCATE TABLE.

At the same time, the NoSQL complaints and problems seem to be presented as very general thoughts, which isn't good. The various NoSQL flavors of databases vary widely, and the way you look at a columnar or graph database is much different than a document database. Really, the piece ought to be separated to look at a certain class of NoSQL database compared with RDBMSes.

That being said, I do see less hype around NoSQL in general. The class of problems handled by various types of databases vary, and I am sure that lots of people using NoSQL databases are quite pleased with them. At the same time, I'm not sure a lot of the RDBMSes did get replaced or continue to get replaced. RDBMSes have a lot of value and we have a lot of experience, both as developers and administrators.

There might be a shortage of NoSQL (pick a type) administrators for operating the databases. Even in some of the famous companies that use them, I find developers often explaining a RCA (root cause analysis) or the troubleshooting of a particular process and fixing an issue. While that might be a choice that an organization makes, a lot of developers that I know don't want to be supporting systems very often, especially at night and on weekends. A couple of bad incidents usually has developers wondering why there isn't an expert on the Operations staff.

I do think that many of the benefits of different types of NoSQL databases are a bit over-hyped. Not that a graph database can't handle your e-commerce system, but it likely takes a lot of code, and likely a lot of mistakes in getting it to work well. I also feel that some of the downsides of RDBMSes are minimized. They are a pain for application developers to work with, and the SQL language isn't very sensible.

That being said, we have a lot of experience with these databases, and that should not be discounted. Productivity is a big advantage in trying to build new software in a flexible manner. I am all for experimenting with different platforms to understand where they work well, but also learn to use your existing RDBMS platform well. Write better code and you might be surprised just how scalable your system proves itself to be.