For the longest time I've heard about the performance difference between a true SQL INDEX and a DDS generated index in IBMi... or 'on the AS400' as the old timers like to say. But is it true? I know that SQL creates its own special access path... and have heard vague rumblings about how it just does things different... and really all of us programmers should be creating files using DDL rather than the old DDS.

So, as part of the project I'm working on we have to copy data from a bunch of rather large files... so I did some real time comparisons and got some fascinating results.

Timings for the big Fight:

Using a physical file with 141,988,325 records in it I issed a batch copy into a DDS version of the file using both CPYF and SQL INSERT and then created a SQL/DDL version of the file and repeated the exercise. Both times running in a single thread job queue when the machine was quiet.

Copy file with 142 million records on an IBMi DDS compiled physical file:

• DDS –> DDS using CPYF took 19 minutes  (it used 99 seconds processing unit time)
• DDS –> DDS using SQL INSERT took 18 minutes  (it used 93 seconds processing unit time)

Then deleted and recreated the file using DDL (SQL) source code:

This weekend I upgraded an old Turnover 5.4 Change Management installation to the newer V100 version across several LPAR's of a customers IBMi intallation.

As part of the mission to tidy up the Software Change process – I see a lot of objects that are routinely created with various parameters being over-ridden because the defaults don’t fit my customers software development model. Now, the problem with this is that a standard is set and then followed, sometimes blindly, for years. Obviously, any step in a compilation where the programmer has to remember to over-ride the default creation parms is time consuming and error prone.

So, wonder why IBM havent changed the defaults for these commands since the good old AS/400 days:

Could we use CHGCMDDFT to set the following parameters as new defaults:

CRTSRCPF RCDLEN(132)

Increase from old RPG3 standard 92 rcdlen, thru RPGLE standard of 112 to new RPG4/Free standard of 132

CRTPF SIZE(1000000 10000 10) REUSEDLT(*YES)

This size is routinely overridden to *NOMAX (not a good idea) so setting a more realistic default size might be appropriate? 1m records?          

Set Source view in *DEBUG mode to *SOURCE allows developers to debug code and see the code it was compiled with:

CRTRPGMOD DBGVIEW(*SOURCE)

CRTCLMOD DBGVIEW(*SOURCE)

CRTBNDRPG DBGVIEW(*SOURCE)      

Read this today. It's cncise and aligned with my way of thinking:

Our IBM i community has one thing in common. We love the platform. However, we are a divided community, in that we don’t consistently treat the platform as the same thing. To some people, we are still working on the same wonderful AS/400 we had in 1988. For others, we have this amazing IBM i operating system running on the best business servers on the planet – Power Systems. And some of us think we are somewhere in between. We don’t all agree on what this platform is, how to use it in a modern way, or what we should call it. Yet we all love it?

Given that our platform has detractors who wish for it to go away, this disunity is a source of material for them. There are many who claim this platform is nothing but an old AS/400, out of date, and out of place. They work to migrate from this platform to a solution that is inferior, and in the long run costs far more to own. Yet, they manage to encourage people to move away from the platform with FUD about how they must modernize and upgrade. They push companies to move with claims that the user interface is old and outdated, and to modernize they must choose another platform.

IBM-I Change Management Application Review

Today's SCM systems are as much focused on productivity gains for the programmer as they are in "controlling" or managing the source and objects. Anything that can be done to make your programmers more effective and more efficient in their development process is a very worthy investment.

Software configuration management (SCM) is an umbrella term that has come to mean a wide range of automation functions for application development all SCM product features have potential benefits and other functions that may require changes to your business practice.

The Big Three

What follows is a listing of products that provide SCM functions for IBM-i application development.

In my opinion, the three big players are:

• Turnover V100 - SoftLanding Systems (acquired by Unicom in 2010)
• Aldon LMe- Aldon Computer Group (acquired by Rocket Software group in March 2011)
• Implementor -  MKS Inc (acquired by PTC in May 2011)

These vendors are the heavy hitters in the IBM-I arena and all have Eclipse user-interface plug-ins for their solutions, which can integrate with IBM's WDSC and RD-i.

The IBMi I.F.S. (Integrated File System) is a terrific part of the operating system which lets us consolidate our native IBMi data and give access to it from other, different, file systems. It lets us access Native IBMi data from internal and external processes. The IFS is a repository for different types of data that native IBMi users or disparate network clients attached to the IBMI can access. In non-geek this means “it’s a way of using IBMi data like it’s on a windows PC”

Whats good about the IFS?

I love the IFS and it’s a sometimes under-utilized function of the AS400… wait… iSeries… erm.. System-i? Ahh that’s right… it’s the IBMi operating system running on an IBM Power Server. This week. *cough*

You can store and access data from/to different file systems on your IBMI. IFS support is integral to the operating system and you don’t have to purchase any additional software to use it. All us IBMi developers know about the QSYS Library and if we want to look at it from the IFS we simply reference QSYS.LIB which is the IFS way of referring to a library. Files end in .FILE – its not rocket science ;)

Why are so many IBMi developers still stuck in the same application design thought process they had 20 years ago?

Why do so many IBMi developers still refer to IBM-i as AS400 (the old system that we used in the early 1990's)?

Why are some people still writing batch report programs and 5250 green screen enquiries?

Evolving past the boundaries of the 5250 display and the reams of printed paperwork is not a paradigm shift.

I want to display IBMi data in a browser, reports as a PDF or spreadsheet, sharing data via a windows network, allowing access via mobile (Droid, Blackberry, iPhone, etc) browsers to the workforce. A whole new world of information-presentation options. If we mobilize IBMi applications our users will be effective from more locations, making them more responsive to changing business conditions. Virtually everybody in every office in every business corner has a working knowledge of the internet and web-based technologies. Web applications are pervasive across all age groups and cultures.

I’ve been playing with the IBMi Integrated File System over recent weeks. The IFS allows us to use and share data with Windows/Linux clients. It basically shares native IBMi (AS400) data with the windows world so that it looks just like a normal windows network drive.

You may read about AS/400 IFS and QDLS. It can be confusing for people. The QDLS folders were orig created for OfficeVision, a discontinued product, but can (and have been) used for other purposes. These folders are DOS-compatable and therefore use DOS style name (8.3 format). Directories (as sometimes called folders) not under QDLS are Unix compatable - long filenames and case-sensitive.

That sounds complicated but to make it very simple