The ServerSide has posted an interview with Dr. Holly Cummins from IBM who recently gave a presentation on Garbage Collection. She expresses her disagreement with some common beliefs about GC, and gives some recommendations for understanding your verbose GC and using it to tune the GC.

Some quotes from her presentation are highlighted below.

In response to the claim:

Garbage collection does work and causes pauses and the pauses prevent my application from doing work so the shorter the pause the better.

Dr. Cummins answers:

Not true! Even when a garbage collector spends a lot of time paused, application performance may be better.

Dr. Cummins also disagrees with the claim:

OK, I get that the application would go faster if I could tolerate long pauses, but response times are critical for my application so the shorter the pause times, the better off I will be”

She goes on to give some recommendations on adjusting heap size, choosing the right GC policy, and using a toolkit from IBM.

The slides are provided online in a PDF.

The video can be seen, along with accompanying discussion on the Server Side (once everyone gets over her appearance).

Ironically, the very existence of an Enterprise Architect may result in your company’s IT system being anything but innovative and simple. Is it innovative to use AJAX because it’s cool? Is it simple to use EJB’s because your IDE has a nifty wizard for them?

I’m not down on the need for an Enterprise Architect – that is how I would describe myself. Yet, companies need to be really careful when they hire one of these that they don’t end up with architects who are so up with the latest technologies, that they become consumed with using them at every turn, even when not necessary (and I would classify most of the new technologies as unnecessary for the vast majority of projects.)

See my post on the costs and overhead of adopting new technologies. Far too many companies have had their software over architected, never benefiting from it, and in many cases having it re-designed when the next architect is hired.

I don’t disagree with anything in the original post, but I’ve seen this technology abuse so much in this field, and seen first hand how much it costs companies, that I also have to add my word of warning.

So the answer to the question of this post, is most definitely yes, that’s the entire reason for hiring one, however, with a big condition added, which is, the company should not qualify the candidate simply for their knowledge of all the technologies out there. There needs to be a sound and conservative approach to software architecture that will prevent the architect from over complicating the solution.

I haven’t tried this yet, as on my current project we are deeply emmersed in Jasper Reports, but this will be interesting to take a look at.

Eclipse BIRT 2.0 Release

While I’ve been working in Java for the past 7 – 8 years, I definitely do not label myself as a Java -loyalist. I’m an Internet Application loyalist, and I want to do whatever it takes to get the apps done right and done fast. I agree with Andreessen’s statement that Java’s complexity has grown by leaps and bounds. The learning curve has become too steep, and many IT departments are finding it difficult to train an employee in all the technologies needed to go in and make a simple change to a module on their web site. When you have to know Spring, Hibernate, Struts, Tiles, SQL/RDBMS, and make edits within all these technologies in order to add one field per the client, it becomes utterly ridiculous.

It may be fun for us developers, and we love all the separation of the various layers of the application, but it’s no good for the client, and that’s who pays us. So we as Architects, Analysts, Designers, and Developers better come up with something that provides for much faster turnaround time.

The report I’m working on now requires a running count for each line item within a group. Fortunately, Jasper has an easy way of doing this, by simply using $V{groupname_COUNT}, where groupname is the name of your group.