What do enterprises care about most?

An oldie but a goodie: Scalability (will it grow when our business grows?). Although performance seems to take a back seat today, I've been through the process of proving with numbers that a specific combination of technology and design will meet future needs. This is still an issue when presenting MS centric solutions to companies who have relied on big iron for years.

I don't think enterprises consider. They don't have hive minds after all. You really need to break it down at least as much as:

1. Business managers (non-IT management folks).
2. Business users (non-IT doers)
3. IT upper management (over both sysadmins and developers)
4. IT pro managers (the folks who manage the sysadmins/operations/help desk)
5. IT dev managers (the folks who manage the dev team (including business analysts, developers, testers, etc.)
6. IT pros
7. Dev team

Each group has very different, and sometimes conflicting, concerns. Conflating them will give a confused and unhelpful picture. To take a most extreme example, a few years ago there was a company whose very viability was threatened by an ERP implementation gone sour. Some of the technical staff continued to push the project long after it became clear it was doomed. They pushed it because they were building up technical expertise for a future with a different company. The project was a success from their point of view because they developed the skills they desired.

The point is not that developers are evil (I intentionally chose to make an example of the group to which I belong to avoid that interpretation). The point is that if you want to define success for enterprise distributed systems you have to move beyond a simple list of the "ilities" and think about the needs of the various stakeholders.

One that I would care about is:

Standardization: how standards-based is the solution. This then plays into many factors, such as:
- security (hopefully we didn't roll our own)
- ability to hire developers that are already familiar with the technology
- technology relevancy - often devs that build a black box find themselves living with it and watching their real-world skills stagnate, which may cause them to get frustrated and quit, leaving a reduced team holding the baby, speeding the next departure and so on

The importance of "Manageability" can't be understated. Where I work MS gets killed on this one because their products don't scale to the size of a Fortune 50 company, meaning when SQL Server starts to run out of legs or starts failing under stress, you're left having to analyze log dumps... There's this huge manageability gap between where the "perfmon counters stop helping" and "having to process a log dump". Anyone who has ever been told to use ADPlus to diagnose a problem with an ASP.NET app will know what I'm talking about.

Craig Andera has a related article on his wiki where he breaks enterprise apps down into five needs. The ones he hsa that I don’t are scalability and availability.

http://pluralsight.com/wiki/default.aspx/Craig.FiveConsiderationsForLargeScaleSystems

Let me come at this a different way. What about interoperability? Isn't it important to think about "playing well with others"? Not just the other systems within the enterprise, but also the ones of our partners. What about transparency? Similar to manageability, transparency is a system quality that tells you how well things are going (not just when something goes wrong). Let's not forget usability. Many an enterprise system has floundered because the users couldn't seem to make it work. Then there's architectural consistency. How well does this system fit with our strategic platforms (hardware, OS, database, development tools, etc.)? We also need to think about adaptability. Can this system adapt to changing requirements? There are more, but the list is too long already.

The problem here is that we have the wrong mental model. The purpose of this sort of list is to focus our attention on a few (3-5 or at worst 7-9) major categories that sum up the necessary requirements. This allows us to filter out the extraneous and gives an opportunity to drill down into each major area. All the short lists (for enterprise applications) I've seen miss something very important. That suggests that we need to re-orient our thinking about enterprise apps. I'm not sure what the short list ought to look like (yet), but I know the ones we have now don't work.