Tuesday, September 30, 2008

Community day: October 1 - What and why?

This post should have been written a few days ago, but as usual, time is precious and never enough...

As the title says, October 1 was defined as the "community day!"... but what Community you may wonder...? Well, it all started at Adam Gartenberg's Blog. Adam is involved in the IBM Information Management division and suggested we could do some coordinated actions to strength the various communities related to IBM Information Management products. So, the basic idea is that we all participate in any community we're involved on October 1.
This of course can lead to various questions... I've been thinking about this and the biggest question in my mind is this: "What is a community, and why do I care? Do I belong to any community?"

The answers to this would be big, probably boring and subjective, but I can't resist putting here some of my thoughts. In these (let me use a buzzword) Web 2.0 times, a community can be anything. A site, a forum, a newsgroup, a blog, a wiki, a group of people with whom I exchange emails etc. In fact, all Informix users worldwide can consider themselves as part of a community. We may never meet each other, but we probably read the same news, the same blog articles, the same forums, and most important, we share a common interest. We probably share some of our doubts, we eventually find similar solutions etc. There is a vast amount of information lying around in the Internet, our email systems, our corporate intranets etc. And then we have search engines, knowledge base systems etc. All this didn't materialize from nothing. It's the direct result of our participation in the community. And when I say "our" I don't mean just the people who write boring articles like this :) I'm also thinking about all the people that simply ask some questions or make some comments.
And now what about the "should I care" part? Well, to put it simply, yes, we should care, simply because it makes our job easier. I can never forget my first steps as an Internet user. I was doing some system/network administration and I had a problem with what we called ATS (Asynchronous Terminal Servers), which were simply network connected equipments with several RS232 ports that connected dumb VT100 terminals to the central system (running Informix Turbo v4 ;) ). My local supplier couldn't find an answer, and using Usenet news I got some help from some friendly guy from the USA. Was this a community? Yes. We both used the same kind of equipment, we had the same problem, and the guy from across the Atlantic had a solution!
Currently, besides my work with Informix, I try to help a customer with other IBM products (namely WebSphere Application Server and IBM Information Server ). As you can imagine I am not a specialist on this products. But I've been able to help this customer in issues and doubts. How, you may ask? Obviously taking advantage of the "community". This include the fine technical support staff, the contents (presentations and other documents) produced by product specialists, the internal and external forums etc. What does it translate into? Productivity. Without these communities I wouldn't have been able to help my customer.

These were just two examples of the importance of community. In 1994, the communities were smaller, there was less content and the search facilities were much less efficient. But the principles were the same as today.

So, hopefully i've showed why the community is so important today. Assuming you agree, I urge you to follow up on Adam's suggestion, and participate more in your community. You can do this by:

  • Answer some comp.databases.informix question(s)
  • Subscribe the IIUG mailing lists or answer some questions there
  • Leave some comment on one of your favorite blogs (if you don't have one, try checking the list on this page's right side) - for example suggesting topics for articles
  • Ask some question to the community or suggest and idea
  • Spread some links of community sites you visit to your co-workers, colleagues and friends
  • Start your own blog?
  • Do some work on a community wiki
The idea of doing it on October 1, is just to create a wave... Did you ever heard about the butterfly effect and the domino effect? ;)

Tuesday, September 16, 2008

Demos? Yes... we have them

There have been a lot of discussions on public newsgroups about the lack of nice and flashy IBM Informix demos. The truth is, IBM has them, but as so many other stuff it may be hard to find in the IBM website. So I decided to ease up your searches. Please follow this steps:

  1. Access IBM TV
  2. On the right side, there is a list of a lot of media titles. Scroll down until you find several Informix titles which include:

    - Mr. Ambuj Goyal (current IBM Information Management General Manager) message about IDS 11
    - A funny video explaining why you don't see much of Informix
    - A flash demo of IDS 11 (overview)
    - A flash demo of IDS 11 security features
I think there is also another flash demo about IDS high availability. I've seen it in some meetings, but apparently it didn't reach the IBM TV website, which by the way is a terrific source of information not only for IBM products but also for current IT trends.
I have the feeling that most Informix DBAs aren't too much into flash demos, but they're nice to show to customers and prospects. So, here they are... take advantage of them as you like.

Face lift

Very short post just to let you know I changed the page look a bit.
There were already too many links, so I tried to organize them into categories (blogs, documentation and just links).

I also added a gadget that allows you to view and listen to some Informix videos available on YouTube.

Monday, September 08, 2008

IDS 11.50xC2: new features

If you follow up some Informix related blogs and sites you probably already noticed that the latest IDS 11.50xC2 includes some surprises. Both Guy Bowerman and Eric Herbert mentioned it on their webpages.
I'd like to add a few more comments to their posts. I chose to do this, specially because the release notes don't emphasize enough all the changes. If you're on Windows, please refer to Guy's post, because it has a lot of features specific for Windows environment.
The features I'd like to mention are:

  • Installer improvements in Mac OS X
    The new installer can now configure some kernel parameters which are needed to take full advantage of IDS. Just recently a fellow IIUG mailing list user was asking about some issues. He tried the IDS 11.50.xC1 installation and hit some issues. After repeating it with IDS 11.50.xC2 everything was ok with the default instance setup
  • Reconfiguration of the Connection Manager without restart
    The new version of Connection Manager allows it's reconfiguration without restart. Just use the -r option. This is a nice feature, since the purpose of the Connection Manager is to collaborate in the increase of the availability of your cluster. So the need to restart it to change it's configuration was annoying. Nevertheless, you should always use more than one instance of the connection manager, so if you need to change it's configuration you should change one instance at a time. Doing this allows you to keep you connection infra-structure answering requests while you go through your CM instances, changing it's configuration
  • Several improvements in the Enterprise Replication facility
    You can check Guy's post and the release notes to check all the changes. They are well covered there
  • Limiting the number of sessions that can connect to IDS
    Once in a while a customer asks me if we have any way to control the maximum number of sessions that can connect to one instance... I usually ask why, and if they can control it on the application side. Sometimes I wonder I they need this, and sometimes they remember me that the most famous IDS competitor has this "feature". I even saw situations where they need to recycle an instance to increase it... Well, I must confess that I've seen situations (specially with Java applications using application servers and connection pools) where the application starts to fire an incredible number of connections (sometimes because there are problems in the database server). In these situations it's a good idea to have a limit of sessions. And after 11.50.xC2 we have a new parameter that allows this. It's called LIMITNUMSESSIONS. If you don't define it, the server ignores the limit. If you define it then the server only allows the specified number of sessions. After that number is reached the new connections will be refused and an error code (-25571) received.
    The best aspect of this feature is that (as many new features) it allows for easy online reconfiguration using onmode -wm and onmode -wf
    Also note that any DBSA user will be allowed to connect even if the maximum connections that we specify was already reached
  • New format for backup/restore filters
    The relatively new feature that allows us to put a whole backup through a custom filter program (BACKUP_FILTER and RESTORE_FILTER parameters) were changed to use another format for backup. This is pretty transparent to the user/dba, but you have to keep in mind that you can't restore a backup made in previous versions with this new release (and vice-versa). So the real impact of this is that you must keep a record of which version you used to make your backups if you use this functionality
  • New BTSCANNER option to specify the compression level
    There is a new option in the BTSCANNER parameter that allows us to define the level of data on the pages that trigger the compression of index pages. This can also be changed dynamically using onmode -C compression and an SQL API command SET INDEX COMPRESSION
  • Server specific audit configuration file
    I recently wrote in the Audit and Role Separation article that one of the problems was that the changes we made using the onaudit utility were written to an adtcfg. file but the instance alwasy looked at the adtcfg file. Well, finally it's fixed, and now the changes and the server look at the adtcfg. file
    Well, if you already played with IDS 11.50.xC1 and redirected writes you probably know that the parameter to activate the redirected writes was called REDIRECTED_WRITES. Well, the parameter changed it's name. It's now called UPDATABLE_SECONDARY. This is not clearly stated in the release notes (although it's mentioned in the documentation notes). Eric Herbert and I hit this issue, so please be warned if you're going to play with redirected writes
    Another feature that may be very useful and it's not well explained in the release notes. With this release, you can use the COMMITTED READ and the LAST COMMITTED isolation level on the HDR and RSS servers. This is a big change, since we all expect these servers to accept only DIRTY READ. Also, these isolation levels were already supported in SDS servers. The only thing you have to do it to activate the redirected writes. This will start a component called the lock manager that is needed to support these two isolation levels. So things like onmode -k start to make sense and be useful on secondary servers.
    Obviously the default isolation level is still the dirty read (even with redirecte writes active)
  • Other features...
    There are several other features, like improvements in the Open Admin Tool, longer (32K) return strings from several string related functions and subquery support in UPDATE and DELETE statements
To completely check all the features I recommend the following links:
So, if you're about to try this version, please check all this information, and enjoy the new features