Simon Woodward - Mar 13, 2008 10:00pm

There are certain things that we take as gospel when we’re building products to run in companies.  Are they true?

* Clients have enormous trouble getting ports opened in firewalls.  If you’re building a product that needs to talk over a firewall then only one port can be used - and even that’s one too many.

>> yup, this is difficult for a number of reasons, especially if its for any vendor initiated remote access (in which case it’s prohibted). If vendors want remote support capability then its actually a lot easier for us to set them up as a contractor, give them a laptop and access to our VPN (there are some maintenance issues associated to this route).

* You never get to sell the product to the end user.  The person who approves the sale and chooses the product is usually a high-up who’ll never use the product in anger.  How much influence do the people on the ground have in purchasing decisions?

>> With a product like Foundation I believe you are looking at a top-down sale (infrastructure director or senior manager), these are not hands on ops people but are technically literate and may well have an Ops background (ie they will respond best to a process/cost message but can understand technical concepts). In my environment the technical resources are brought in to verify operational fit.

Also, there are some things that we as engineers just don’t know.  Any answers?

* How much do the end users actually tend to use these expensive tools they buy?

>>that entirely depends on how well the tool is embedded into the organisations operational culture/processess.

* How accurate are the CMDBs and asset databases in real-world companies?  Do they generally have automated scanning systems feeding them?

>>again, it varies - it is rare to see a good end to end story and rarer still for a solution to exist across the entire assetbase, in my experience the one size fits all approach to the structure of a CI record leads to too many compromises (thats part of the reason why I like the sound of Tipple)

* What use is made of the data in the systems, and how often?

>>if other IT people learn that you have good quality inventory data then they will exploit it tactically, getting your data used on a strategic planning level or getting it embedded in operational process so its working really hard for you requires cultural/organisational realignment is much more difficult.

(edited for layout etc)

Simon Woodward - Mar 13, 2008 10:00pm

* Clients have enormous trouble getting ports opened in firewalls.  If you’re building a product that needs to talk over a firewall then only one port can be used - and even that’s one too many.?

True. I reckon you mean both internal firewalls (DMZs) as well as outer firewalls.
Outer firewalls are often impossible to open. In my opinion the troubles related to getting ports open depend on two variables (just like most changes):

1) the (desired)level of security
2) the level of organisation

When 1 is high and 2 is high it probably will be medium trouble.
When 1 is low and 2 is high, in general it doesn’t take much effort.
When 1 is low and 2 is low, one can probaly sqeeze it through quickly as well
When 1 is high and 2 is low, one can wait forever

Unfortunately many companies fall in the ‘1 high/2 low’ range.

Simon Woodward - Mar 13, 2008 10:00pm

Also, there are some things that we as engineers just don’t know.  Any answers?

* How much do the end users actually tend to use these expensive tools they buy?

That depends on:

1)the availability of other tools
2)the effort the end-user has to put in (the usability).
3)the quality of the tool/data
4)the extent end-users have been involved in the project.
5)the pressure from the management to use the new tool

My experience is that if other tools are made unavailable, the end-user often turns to the new tool ;-) But, often the quality of the new tool is not enough to take over from the old tool right-away.

Simon Woodward - Mar 13, 2008 10:00pm

* How accurate are the CMDBs and asset databases in real-world companies?  Do they generally have automated scanning systems feeding them?

I’ve been working for several big companies now and the best data is always in obscure excel sheets and well defended hostsfiles etc. CMDBs always run behind and are often incomplete because they take too much time to maintain.