jump to navigation

Friday Philosophy – Sometime The Solution Has To Not Only Match The Problem But Also… August 4, 2017

Posted by mwidlake in Architecture, development, Friday Philosophy, Perceptions, Programming.
Tags: , , ,
trackback

…The People!

When you design a system for end users, a good designer/developer considers the “UX” – User eXperience. The system has to be acceptable to the end user. This is often expressed as “easy to use” or “fun” or “Quick”. But in reality, the system can fail in all sort of ways but still be a success if the end user gets something out of using it. I’ve said it before and I’ll say it again and again until I give up on this career. In my opinion:

User Acceptance is the number one aim of any I.T. system.

OK, you all know about UX probably. But what about solutions that have no End Users? I’m thinking about when you create a technical solution or fix for an internal system, to be used by fellow I.T. professionals. How many have you considered the skills and temperament of the people who are going to house-keep the solution you create? I suppose I have had opportunity to think about this more than some of you due to how I work:- I’m a consultant who gets called in to fix things and then leave. At times I have chosen a solution that has been influenced by the people who will be looking after it.

I’ll give you an example. At one site that I worked at for about 9 months, I did a lot of work for one system. The developer/systems administrator who looked after the system was…stupid. I don’t really like saying that, we all vary in our skill set, experience, intelligence, *type* of intelligence (I know some people who can speak 3 languages or know a lot about history but could not wire a plug). But this guy really seemed to struggle with logic, cause-and-effect or learning anything new. And I had to help him look after this database application with one main, huge, hulking table. It had to be partitioned, those partitions maintained and the data archived. I implemented the partitioning, I explained partitions to him several times, what was needed to maintain them, where to look in the data dictionary for information. It was like talking to my mum about it. He just seemed not to understand and his efforts to code something to do what needed to be done were woeful.

I knew it was not me, I’ve run enough training sessions and presented so often that I know I can explain myself (Well, I hope so! Maybe I am deluded). He just was not getting it. Maybe he was in the wrong job. So I wrote him a set of SQL-generating scripts to get him going. He kept messing up running them. In the end, I knew I was about to leave and when I did within 3 months the real customer would have a broken system. So I wrote a mini-application in PL/SQL for him to do what needed to be done. And set it to email a central team if it failed. The team he would call when he broke it all again. I also simplified the solution. My original system had some bells and whistles to help with future changes, such as over-riding where new partitions went or how old ones were compressed. I stripped it out to keep it as simple as possible. I altered the solution to suit the person that would run it.

I’ve done something like this a few times over the years. Usually it is more to do with the skill set of the team as opposed to actual ability. I’ve on occasion worked with people who are new to Oracle and my time is limited so, rather than give them a solution written in PL/SQL that none of them know, I have done so with SQL and cookery instructions/shell scripts. It’s not the best solution but it is something they can live with.

More recently I had to look at fixing the performance of some SQL statements. Baselines would have done the job perfectly. However, the team were all Java experts and had no desire at all to learn about database administration. (To be frank, they had no time to learn either, it was the usual situation of them having 75 hours of work each every week as management thought just shouting would get things fixed, not hiring enough people). I strongly suspected that they would forget about the baselines and if they had a problem they would be confused as to what was going on. So I fixed the key SQL statements with a set of hints to force both the overall structure of the execution plans as well as which indexes to use etc – and said over and over and over and over that if they ever changed indexes or migrated to a later version of Oracle, those hints would need reviewing. They were, in effect, part of their code base. A big advantage of the hints was that they would see them in their code and it would remind them what had been done. They seemed happy with that.

My point is, sometimes the “best” solution is not the correct one, even when you are keeping within the walls of the computing department(s). Sometimes you need to think about who you are giving the solution to and change the solution accordingly.

Advertisements

Comments»

1. jgarry - August 4, 2017

We Are All Users

Best Solution !≅ Best Technical Solution

2. stewashton - August 6, 2017

I remember the time I was asked to simply lessen the time taken by a batch process. I estimated it needed to be 10 times faster to fit into their batch window if volumes increased as expected. I replaced a long Java program with one SQL statement that ran 90 times faster.

Job done?

No. Management said the developers would not be able to maintain the SQL statement.

This was perhaps more a problem of MX (Manager eXperience) than UX.

I realized later that if I had broken the SQL statement into parts, with intermediate results held in temporary tables, and put Java glue around them to run them in the right order, I would have done the same work in almost the same time in a perfectly acceptable manner. The problem was the total absence of Java in the solution.

One year later the same managers contacted my same department to ask for help with their too-slow batch process.

Non-acceptance can sometimes be due to perceived problems, not real ones.

Best regards, Stew

mwidlake - August 7, 2017

Absolutely Stew. I find that one of the hardest things about being a consultant called into a site is to get them to accept any of the technical solutions. So often they want a fix that includes no changes to code, no changes to the database, has no drawbacks and is guaranteed to work. For ever :-).
That is partly why sometimes the solution you give them has to be design to fit their limitations, even if those limitations make no sense. If I can’t shift their opinion at that time but I can give them some sort of solution, well I do that. Hopefully, if there is a next time, they might either be willing to trust me more or had enough time to adjust their stance on a couple of things.
Letting it all go horribly wrong by giving them no solution occasionally works!


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: