Informix … and marketing.

Odd times out there for Informix. IBM is currently doing an “11.7 Technical Roadshow” around the country. The Informix twitter account posts once or twice a day. The name “IDS” is now just “Informix”, and hasn’t included “DB2” in the name for quite some time now. I don’t think it’s even “IBM Informix” anymore, but that depends on what page you visit on the IBM site, etc. I’ve been around the Informix world since 1995 – and I’ve seen the good, the bad, and of course, the ugly. What would I like to see now? The Informix community to recognize what is being done to promote the product. Is it enough? It depends on who you ask. But the gloom and doom chatter out there is getting real old in my opinion. To say “nothing is being done” is completely inaccurate, and seriously unfair to the core Informix crew at IBM that IS working hard to get the word out. Folks – “we” missed our window a long time ago. It started with our own “mini-Enron” long before the acquisition. Then it worsened with the “Gold Bundles” that were supposed to bring all the Informix clients over to DB2. But starting w/ IDS 10, we did a “World Tour” (where I was the technical speaker for the sales team across numerous countries – it about killed me. 😉 ), 3 major releases have come out since 10 (11.1, 11.5, and 11.7) and a brand new architecture in the Informix Warehouse Accelerator, and a number of huge features brought over from XPS.

So – am I sad, angry, or disturbed that our “Informix world” may never claim it’s deserved place in the industry – of course. I think I’m more angry that “we” won’t all be able to retire doing Informix work. But (to be original) it is what it is.

What confuses me the most are the “gloom and doom” advocates – don’t they realize that their constant negative comments and complaining on CDI and other places do not help, but actually promote the negative impression even faster? What part of that is confusing? Folks – all I want is accuracy, and not one-sided, “all negative” comments. If you have ideas for helping promote – send them on. And remember – you CAN prop up Informix with good work like “Case Studies”, “Success Stories”, and the like. And those of you that know me well – you KNOW I am always trying to be honest, up-front, direct, etc. So save the feedback that I don’t want ANY negative truths to be spoken. But man – to ONLY post negative stuff? That’s just hurting the cause of Informix overall. And like I said – it’s getting really old to hear the same complaints. I have been busy doing Informix work since 1995, and it hasn’t stopped yet. I doubt that I’ll do Informix work forever, sadly enough. But I’m going to enjoy the ride while it’s here.

Informix 4GL … and some say it’s “gone…”

Received a call from a distributor asking if I did 4GL education, to which of course I said “sorry – I don’t.” Never really had a reason to learn it over all these years. I commented to him that “many people say it’s no longer around.” He laughed and said his company has a couple thousand clients still using it very strong. I agreed – I’ve seen a ton of job openings – both contract and permanent – for Informix 4GL folks. I think 4GL will end up like Cobol … everyone swears it’s gone, but those in the trenches see it all the time.

Random Informix Ramblings…

Almost every client I’ve visited in the past few months is talking about migrating to “10 or 11 of IDS.”  Of course that’s exciting for us “Informix-faithful”, and those of us that have “been around” for a long time. The more talk around migration, the more potential work for us Informix techies….Talking with some of the internal development folks regarding OAT (the Open Admin Tool), rumor has it the Adobe Flex is front-ending lots of “stuff” now, typically over PHP. Rumor also has it that you can plug-in your own set of scripts/tools/etc. into OAT…I recently spent 4 weeks at a client site working on PA-RISC HP-UX and running an application that was at least 10 years old. Small engine of only 20G or so, and there were performance concerns. During the time there, I was able to write a new module or two for my  “IDS Sanity Kit”, including a “Benchmark” module that allows dynamic sql to be used for testing. The “Sanity Kit” has been evolving since the “original Watcher” I wrote circa 2003 for a long-term consulting assignment (while still in the education group though.) We at Xtivia use the Sanity Check module for the standard baseline IDS Health Check most of the time.

Monk’ing IDS

May2011: “Monk” to the Rescue – playing a bit of a “private investigator” for a site that has an apparently random response time for a specific query of 8-10 minutes. Confident we found the issue last week – still testing and eliminating other possibilities. Fun work, even though it can be a bit of “hit and miss” until you start to hone in on a pattern, or on this case, something kept appearing that just “shouldn’t be there during this time of day.”

As it turns out, it was in fact the “rogue job” that wasn’t supposed to be running. And of course, it was an UPDATE STATISTICS job that had crept into a longer and longer job, not finishing nearly when it was supposed to finish. I wasn’t really looking for this as the culprit though. I was told that the client thought for sure it was a “locking issue.” And from what they described, their diagnosis sounded pretty good. So I was heading down that path to capture queries during that time, see who’s waiting on locks, etc… Much easier in 11.x if you fully understand and have used the new sysadmin database and the ph stuff. But since I was under the gun, I was starting to craft my own scripts (as we all have done over the years) to drill down a little a time.

Once I had some scripts in place, I decided to look closer at the historical “Watcher” data I had been collecting, focusing of course on the window of time the client specified. I noticed a few odd things, but nothing really stuck out. Some clients have to run a flavor of UPD STATS during the day, but typically they’re quick-hitters. I noticed a high value of “2” for UPD STATS jobs during the window mentioned. So at most, there were 2 UPD STATS jobs running. That seemed odd(er). I checked the crontab and the UPD STATS logs and the nightly UPD STATS was supposed to kick-off around 2am, and it only typically ran for a couple of hours. The “bad window” was 7-10am though! So for whatever reason, this job had spread out in to the morning work hours. I noticed it had a PDQ of only 50, and the DS_TOTAL_MEMORY was not set very high at all. So the PDQ wasn’t helping too much at all. I had just finished a new “MSTATS” package that generates all needed and up-to-date UPD STATS scripts for an engine, based completely on SQL and shell scripts. I decided to implement the new MSTATS instead of trying to backtrack as to why the old UPD STATS jobs ran longer than normal. (In Art’s defense, we all love do_stats and have for years. I wanted one that didn’t depend on esql/c for now. In John’s defense, the new “Auto Update Stats” will be the Jaguar of UPD STATS at some point, but it only works against logged databases. All these databases were non-logged, so …