“We have already discussed the reliability, cost effectiveness,
and efficiency of databases on an Enterprise (mainframe/midrange)
Linux installation. Today, we will talk about the scaleable aspects
of such a solution. Hopefully by now, you will have had the
opportunity to peruse the http://www.tpc.org (tpc.org) site as I
will be referencing it for today’s discussion.
“A scaleable database means many things, however, I will use the
following definition of database scalability: A database is
scaleable, if and only if, its performance grows linearly (or
better) as it’s size increases. For example, if I have a database
3GB, 300GB, or 3TB in size, my administration, queries (updates,
inserts, deletes, indexing, etc…), and analytics (OLAP, data
mining, statistical analysis, etc…) have linear or better
performance assuming appropriate hardware infrastructure.
“Many good and functional databases are available, but many
saturate and decrease in performance under higher database loads.
Some will even try to use clever caching tricks to make them
perform well, but these do not work well at the 1TB limit or
beyond. Based on this, they would not be scaleable by my
definition…”