macOS DB2 Beta Released for OS X

Don't expect this to be considered a major news item, but for anyone interested in databases, and there should be a few on this forum, IBM have (finally and long last after many years of hints and suggestions) released 9.5.2 Express-C for OS X.

It is 64-Bit, Intel Leopard only. Installation doesn't seem straightforward at the moment, certainly not up to Linux or Windows installers (haven't installed yet, hope to sort it out over Christmas break). Don't expect anything along the lines of the MySQL ease of use and set-up for the time being, it is very much along the lines of "OS X is UNIX", not "OS X is a user-friendly OS".

This is the free version and currently a beta release - it is only really good for development and only time will tell if IBM go the full server route. I am just happy to see tangible progress made to support OS X as development platform.

Anyway, thought people might be interested.

http://www-01.ibm.com/software/data/db2/express/download.html?S_CMP=ecddww01&S_TACT=mac952

Well, installation is not without a few hiccups, you need to create or edit /etc/sysctl.conf with this content:

kern.sysv.shmmax=1073741824
kern.sysv.shmmin=1
kern.sysv.shmmni=4096
kern.sysv.shmseg=32
kern.sysv.shmall=1179648
kern.maxfilesperproc=65536
kern.maxfiles=65536

It isn't entirely clear which tools are included - as far as I can see it is simply the database server, and not the administration server or management client software. This is clearly a beta release.

On the other hand, this is more than enough to build and develop with, especially since deployment will be on Windows / Linux. I will try and get some stuff running on this over the next few weeks and see how it fairs with some data, especially compared to parallels.

(There is a thread on this on the DB2 support forums, apparently more documentation will follow).

For me it has been running fine. I am using it to store sampled data so I can work away from the office to a certain extent, and almost exclusively in Java / shell.

I haven't used to CLI from c/c++, PHP, Ruby or Python bindings (I believe these are all working though). I would like to see if there is a way to get an ODBC connection working for Excel but I have been too busy to look into this.

More than anything, I am aggregating and sorting data. In terms of general performance, my core data sets range between 5 and 10 million records, which are grouped and aggregated by an unindexed field in ~20 seconds (with Java data IDE and Java development IDE running on my work 2.x Ghz MBP with 2Gb RAM). Joining the same query to a dimension table with ~5 million records drives this up to ~5 mins (depends on indexes, result set ~0.5m records). All the olap functions I use on a regular basis are present.

Downside - there doesn't seem to be any sort of real management software (as on Windows and other Unix) so all table stats need to be generated manually / cron-ed, same goes for data transfer. I needed to create a launchd job to start it off on booting.

Defaults parameters needed tweaking (log file size, buffer pools etc.) and almost certainly some performance improvements can be found but I am not going to spend too much time DBA-ing.

It is miles better than SQL Server (or anything) on Parallels, though, which tended to balk at anything more than a few K rows (obviously disk access horrendous). I am not holding out hope for MS to port SQLS to the Mac ... although they would make me very happy if they did.

It has been a while since I tried to get MySQL doing anything similar (was very unhappy with anything 1 million rows and really seems happier as a web DB rather than for any sort of number crunching not to mention the weird syntax, although i confess I know much less about improving MySQL performance) and have never tried Oracle on OS X - I was rather under the impression it had been largely abandoned and is the about the one DBMS we don't have at work.

Overall, there is quite a bit of overhead but it by far the best DB I have found, for me, on a Mac.

We only have Postgres in some web apps but the just released 8.4 http://www.postgresql.org/about/news.1108 looks like it has a very good range of olap functionality. From my point of view, it is really a matter of time and compatibility with the applicable servers. That and it takes years to get to really know a dbms.

DB2 comes in several versions:

Express-C is the free version, limited to RAM and CPUs, and the version available on OS X.
Express, Workgroup and Enterprise allow increasing scalability and features for HADR and management software. I forget exactly what is included with what. The is also a Data Warehouse version, including data mining and a whole host of analysis / BI stuff (a bit like MS Analysis Services) and development tools (presumably to suck you into buying the pricey Rational tools).

It depend on which version suits your needs.

Pricing runs the full gamut, too Spend as little or as much you like.

OS X is not supported as a server ... I am sure IBM will sell you a z-series, though

There are some good introductory postings on http://antoniocangiano.com/ and he was, until recently, a Mac user too.