Thursday, September 14, 2006

Do you feel insecure?

Security. Love it or loathe it? Most of the time the choice is yours, you can choose not to lock your house or car, you can choose not to fly, you can choose to keep your PIN in your phone memory under PIN, what you probably can't choose is whether or not to apply it to your Analytics system, because some one else has made that decision for you and they pay your salary!

A brief scan of the popular analytics message boards does suggest that this is part of the system a good number of people can't get their heads round, so I thought I'd jot a few of my thoughts on the subject.

Within Siebel Analytics there are two basic types of security, Data Level and Object Level.
Actually, there are EIGHT layers of security in an Integrated Analytics application (can you identify them all?).

Data level is about controlling the data i.e. the facts that a user can see, typically this is because managers don't want fisticuffs in the office; so preventing everyone from seeing how much commission the sales team get is generally a good idea. Also, you may be familiar with TMI, Too Much Information, if like me you're a nosey parker than everyone else's data is far more interesting than your own and you can't resist having a look should the opportunity present its self. Giving all your staff this opportunity can have a dramatic effect on productivity.
On a serious note in the financial markets you'll know all about 'chinese walls' and the need to keep data secret.

At the simplest level, this is how to go about administering data level security;
  • In the Analytics Administration Tool go to Manage>Security to open the Security Manager, the security objects are displayed down the left,
  • Click on Groups and then right click in the r.hand pane and,
  • Select New Security Group, the procedure is the same for adding a new user.

At this point you should already have thought about whether you are going to administer security at group level or individual level, this will be largely dictated by the number of users, again it is probably not your decision to make so I hope the chiefs in your organisation know what they are on about!

So, for example you have created a group called Agents, these could be support staff or telesales, it doesn't really matter, the point is you want to limit their view of the data to those applicable to their geographical region of responsibility. In the right hand pane double click the Agents group, and click the permissions button and select the Filters tab, now click the Add button to add an object (i.e. table) whose content (data) you wish to restrict access to, highlight the table and click Select, now check that the Status is set to Enable and click the Ellipsis button to open the Expression Builder, the world is now your SQL oyster and you can do pretty much what you want to restrict the view of the data available from your chosen table, and like oysters, SQL should be chosen with care to avoid indigestion!

The choice with this type of filter is between hard coding a value, so the filter will always filter the same value, this will seem like a bad idea if the data is changed, for instance if the Regions are renamed at some time in the future. A better solution is to use a variable that returns the values available for the specific user or group at the time of logging in.

Anyway, it's late and there's so much more to this subject that I will deliver it in small bite sized chunks.

To be continued ......

PS. The eight layers...

  1. Application Log in (Application Authentication)
  2. Responsibilty to a View
  3. Access to Analytics system (Analytics Authentication)
  4. Access to Analytics Components (privileges)
  5. Rights to Webcat object (e.g. Particular Dashboard, Report, Folder)
  6. Rights to Analytics Repository Object (Usually set on Presentation Layer)
  7. Data Restrictions (via Security )
  8. Database Access (normally uses shared name in connection pool)

No comments:

The Cowes

The Cowes
Cowes Racing