Show Posts

This section allows you to view all posts made by this member. Note that you can only see posts made in areas you currently have access to.

Messages - kepNCI

Pages: 1 ... 4 5 [6] 7 8 9
Uml Process / Tracing network device locations
« on: February 22, 2015, 01:30:15 am »
Looking for suggestions on the best UML elements and connections for representing geographical locations and relationships with network devices in such a way that one can trace the location of a given network device.

For the network devices, I already have a package and deployment diagram for each type of network device types (server, router, switch, etc.) as well as stereotypes defined in an MDG for each type from which I create instances for each network device in the respective deployment diagram. Most of my network devices have a tag for location, which is populated from values obtained from our discovery tools. However the goal is to have a complete traceable relationship to its location, which I believe implies some relationship connections as well as package hierarchy for locations in browser.  

This is what I have tried so far:
I created another package called Facility Locations and an associated Facility Locations diagram. Under the Facility Locations package, I have a package for each country. Under country I have a package for each location/city, then under city I have a package for building. Thus in the project browser one can see the geographical country, city/location, building hierarchy of where we own/manage network devices. Then in the Facility Locations diagram, I basically duplicate the same hierarchy structure, by dragging each geographical package as a package element and then connecting them appropriately using a nesting connection.

Then under the package for each building I have a deployment diagram onto which I drag links to the device instances that are located in the building. To create a traceable link between device and building, I also dragged the associated package for the building as a package element onto the diagram and link each device using a nested connection.

Now when I click on a network device, I can trace its location within the traceability window.

- Can you think of a better way that may be more UML compliant?
- Are package elements the best element to use for defining a geographical location?
- Is the nesting connection the best relationship connection?  Seems to be the only choice for connecting the package elements.
- Is the nesting connection the best relationship connection to associate the device node to the building package element? When you use the quick link from the node to the package, nesting connection is not an option, however, EA allows one to use the nesting connection from the toolbox to create the link.  Does that mean I am not UML compliant?

Uml Process / Re: Capturing port details on switches, routers, e
« on: November 06, 2013, 12:49:11 am »
Thanks, Andy.

In this case, I do not think Run State helps me any.  

However it did get me looking at another issue, that is, how I could designate a network instance as a future or proposed deployment. From what I could read, Obejct state would even be better, but for some reason it is not an option on the Advanced memu for my node stereotyped instance.

Uml Process / Re: Capturing port details on switches, routers, e
« on: November 01, 2013, 08:05:50 am »
I'm with you, i.e., I decided to use ports. I am having problems creating instances of the defined switch model after I added ports.

SETUP: To defined the configured switch:
I have a node stereotype defined for a Switch in a toolbox, complete with associated tagged values (manufacturer, model, # ports, serail#, location, IP Addess, etc).
In a package/diagram called "Switch Types". I have dragged the Switch node from my toolbox onto the diagram and call it "Switch" to represent a generic switch.
I then drag the same Switch Node onto the diagram and called it "Cisco" to represent a generic Cisco switch and created a generalization connection to "Switch" node. I set the Manufacturer tag to "Cisco" and delete all of the other tags (allowing the others to be inherited from the parent "Switch".  
I then drag the same Switch Node onto the diagram for each of the Cisco model switches we have and make a generaalization assoiciation to the "Cisco" switch, updating its tag for Model and # ports accordingly and deleteing all other tags (allowing it to inherit the Manufacter from the parent Cisco and the other tags from the grandparent "Switch".
When I want to create an instance of the specific model (I have another package/diagram for all of the switch instances, I drag the model node from the project browser onto the diagram and state I want an instance, then update the tags for the instance accordingly, deleting any null tags, allowing them to inherit from parents.

This method has worked great so far. But now I have a probelm.

On the Cisco Model node in my package/diagram for Switch Types, I added a port per the switch configuration.  Now when I drag that node onto my Switch Instance Package/Daigram, it allows be to select the embedded components and I select all.  Probelm is that it appears to create an instance of the port on the new switch instance, but it has no port name and some connection back to the classifier's port.  For example on the Switch Model config node which would be the classifier I have a port called "1/1".   When I create an instance with embedded ports, I get a port which in unnamed, but labeled on the diagram as "/1/1" which apparently is a reference to the classifier port. If I update the name on the instance to "1/1", the diagram label shows "1/48 /1/48".  What I think I want is an instance created where the ports are also instances with no classification back to the parent classifier, but cannot figure out how to do this, unless I create a new ports of the instance.  

Hope the above makes sense.

Uml Process / Capturing port details on switches, routers, etc.
« on: October 31, 2013, 05:18:24 am »
I am using EA to document our production network config, mainly for the purpose of future design. I want to capture the some port details on switches, routers, etc. I know UML does not offer any guidelines for network configuration. Visually on the diagram I see two ways of doing it and was wondering if anyone has any experience and/or suggestions as to which would be better for not only the visual diagram, but also for reporting/documentation.
1) actually link a port stereotype on the node and make the communication connection to the port.
2) do not use the port, but instead make the communication connection to the node, but then use the communication connection's source/target qualifier properties to name the port.

Uml Process / Re: UML relationship between cluster and server
« on: February 27, 2013, 09:43:50 am »
I think you are right.

It is more an academic issue in my head as I try to learn UML than an important issue for our diagram and project since the main thing we want to show is the relationship.  It is just that I want to try to stick with correct UML, if possible, but since this whole network config is somewhat outside of UML, it likely does not matter.

Uml Process / Re: UML relationship between cluster and server
« on: February 27, 2013, 03:34:31 am »
You hit right on why in my case I likely should not use composition.  In our network we may have 3 physical servers in which each are associated with three clusters.  Thus deleting one cluster would not necessarily delete the associated servers, or vice versa.  However deleting all three servers would effectively delete all three clusters.

Aggregation may be the best in my case. In which case who owns who? In other words, Is the physical server the owner or is the logical cluster the owner? In my case owner would not necessarily imply life dependency.

Thanks for the reply. Got me thinking.

Uml Process / UML relationship between cluster and server
« on: February 21, 2013, 04:59:04 am »
I have defined our network devices, including servers and clusters.
I now want to create a relationship between a cluster and the servers that make up the cluster.  Which UML relationship best describes the relationship between a cluster and its servers? I am thinking just an association connection.


Uml Process / Re: node vs deivce (EA bug or UML restriction)
« on: September 28, 2012, 04:39:17 am »
BTW: The Advanced/Instance Classifier... appears to be working the same as the Parent... function, that is, device stereotypes are not displayed as an option.    

Uml Process / node vs deivce (EA bug or UML restriction)
« on: September 28, 2012, 04:31:14 am »
I created a profile for generating network deployment diagrams. I created device stereotypes for all of the hardware including servers. This seems to be in agreement of what I read about devices in that it is just a stereotype for a node which represents a physical piece of hardware.

That being the case, why when I use the Advanced/Parnet... function only stereotypes derived from node are displayed as a choice. None of the device stereotypes are selectable. Is this an EA bug or is there a UNL rule that prohibits a device from being a parent?


Uml Process / Re: Icon in Node
« on: August 30, 2012, 11:59:46 pm »
Thanks, Roy

Will play with that today. I have not done much with scripting, yet.

Will I run into the same issue with a shape script decoratation as f I do with using a shape script to assign an alternate image to a stereotype, that is, when I use the stereotype with a shape script image, the Feature Visibility option to display the tags and notes does not work.


Uml Process / Icon in Node
« on: August 30, 2012, 04:44:40 am »
Some of the EA supplied stereotypes for a Node have a small icon in the upper right hand corner of the node rectancle.
Are these icons defined by UML (as in the case of the component icon)?
Is there a way to add your own icon to node and device stereotypes that are user created?
(if so, would hope that it also works for stereotypes defined within an MDG profile).

BTW, q
I confirmed your observation that it appears that the initial default values and order for Project Difficulty, Priority and Test Status are hard coded. If changed via the menu option, including the order, then t_list is used. Value of Category reflects the list, that is:
- ConstStatusType
- DifficultyType
- PriorityType
- TestStatusType
NVal denotes the display order
Do not know how the default is maintained. There is no flag in the table row and Notes is null.


Is EA 10 compatible with EA 9 database?
Can EA 10 and EA 9 co-exist on the same PC? (Although I would not want to use EA 10 to access an active EA 9 project).

Just some observations:

Appears to be a table set up to create various control lists, but looks like it is only being used for Settings/Project Type/General Types/Constraint Status Types.

Records Changes, Defects, Issues or Tasks related to element. Managed via View/More Element Tools/Maintenance

Allocates time for specific resource on this element.  (see View/More Element Tools/Project Management).

Tests defined for element via View/More Element Tools/Testing

Cannot find source in database for values maintained by:
Settings/Project Type/General Types/Difficulty
Settings/Project Type/General Types/Priority
Settings/Project Type/General Types/Test Status
Settings/Project Type/Estimation Factors/Default Hour Rate

BTW.  As suggested, I did create a data model of EA database via re-engineering and added notes to many of the table based upon observation of the type of data contained in the tables as well as added the apparent relationships of the tables that would be most used in queries. Helps quite a bit.


Just bought/downloaded your book.  Looking forward to reading it.


Pages: 1 ... 4 5 [6] 7 8 9