This article is written in English and Portuguese
Este artigo está escrito em Inglês e Português
I recently got an advice on how to make better use of Twitter… And so I did… I increased the number of people or accounts I’m following and today I was flooded by messages about the launch of one of the database competitors. If you’ve been paying attention to the net you probably know which one I’m talking about… I’ve seen some references before and I decided to investigate a little further what were the new features causing all this buzz around it… I must grant credit to the company behind it, since it was not difficult to get information and find several articles and papers and people talking about it.
I think that in the IT field we tend to close ourselves around what we know better. I’ve seen it in Oracle DBAs, in people working with Informix (you should know it’s much harder to find an Informix DBA than a DBA from any other database since we tend to have several hats and play several roles) and with people working in different environments (z/OS is a classical example). And apparently it also happens with people working with SQL Server. I’d say that only this can explain all this enthusiasm… Let me explain why, by picking two of the flagship features of it’s new version (I’ll be using the terms I’ve found on the Internet in blogs, articles and so on):
Believe it or not this is a form of replication that allows databases to be put together in groups that use a primary server and one or more secondary servers. The replication can be synchronous, or asynchronous. The secondary servers can accept read only queries. It includes some sort of connection redirection and something I could not exactly understand that allows temporary statistics to be computed on the secondaries and stored in temporary spaces…
- ColumnStore indexes
This is interesting…. It combines several technologies like in-memory database, columnar storage and star model optimization. This allows incredible time savings, but has some drawbacks, like not being able to update a table with an index of this type (several workaround are mentioned, but all of them have serious implications). It’s up to the optimizer to decide if it will use this kind of index or the traditional query plans.
I’m sure that if you’re an Informix user, or someone paying attention to the Informix scene, you have a smile on your face by now… And I would not need to explain why. But for the people who are a bit more distracted, or as I mentioned above live on a closed world, let me explain why Informix people have a smile on their faces at this point in the article:
In 2007 (yes, five years ago), IBM introduced version 11.1, code named Chetah, and one of the features was something called MACH-11. This did not cause half the buzz that we’re seeing today, but in very short words, it was the ability to configure a set of Informix instances (where we can have several databases) with a primary server, a “close” secondary server, called HDR (which can by synchronous), and several remote secondary servers (RSS) which receive the logs. The communication between these can be encrypted (there are known customer c…