please most readers of this book. However, this also means that, unlike
with most enterprise-focused relational databases, there is probably not
any enterprise support available for businesses. This will frighten many
managers.
Most NoSQL projects are fairly new. This also means that they are
untested in large enterprises and businesses. It can also mean that it takes
longer to set up because of the learning curve; known solutions can be
implemented more quickly.
NoSQL is not always ACID (atomicity, consistency, isolation, durability)
compliant. Instead of guaranteeing that every data transaction is instantly
and properly recorded, that only one interaction may occur with a piece
of data at any given time, and that only the current version is available to
the end user, NoSQL databases often work on a system of replication of
data across multiple hosts that each get updated eventually. This may
happen quickly, but there is no guarantee that at a given moment you
will retrieve the most up-to-date data. This matters when dealing with
financial transactions but might not be a concern with web search results
where “close enough” might be all that is wanted or necessary.
There is no particular advantage to NoSQL unless you have data that is
large enough to benefit from it or your project fits neatly into a specific
use case.
It isn’t difficult to migrate to NoSQL from traditional relational databases
if the need arises. You know Donald Knuth’s saying in The Art of
Computer Programming (Vol 1-4A Boxed Set. Published by Pearson
Education) that “premature optimization is the root of all evil.” It might
be wiser to design your site or application using known technology and
then migrate later if the need arises.One facet that is sometimes described as an advantage and at other times as a
disadvantage relates to administration. Relational databases often must be
administered by trained staff. The positive side is that qualified database
administrators (DBAs) are plentiful, if expensive. NoSQL databases are
designed to be created and, at least in theory, require little to no further
maintenance. Some pundits claim cost savings; others claim that DBAs will
still be needed (though for different tasks) and that there are very few people
who are trained and available to perform those tasks because of this
perspective and the newness of the database style.
NOW, A QUICK WAFFLE ON THE NAME
Google and others seem to like the name “NewSQL” more than “NoSQL.”
Some only apply “NewSQL” to a certain group of databases that are