Ashish Gupta wrote:


As per the sysadmin guide , it is explained as : All the CI that are feeding data into Email are Upstream CI 

That's not completely accurate... the relationships depict dependencies, not data flow.

Similarly how Email servers can be affected by email service , it should be vice-versa , if Email server goes it down , then it will affect Email service

That's more like it.

Here's one way of thinking about it: see the arrows in the diagram? Consider them line-of-sight. Consider people standing on each CI in the diagram, looking in that direction.

Now, if something happened to a particular CI, all those looking at it (upstream, looking down) will become aware - they'll see it, they're the ones affected by it.   But those downstream of the affected CI won't be looking backwards, they'll be blissfully unaware.

Another way is considering change impact.   In your diagram, if I were to make a change to IronMail-SD-02, I can see that this would affect the Business Service: Email, as well as those upstream, so will need to notify those users affected.   However, EXCH-SD-05 and EXCH-SD-07 would be none the wiser, since they're downstream.

One last way is to consider problem diagnosis: if something were reported about a specific CI (IronMail-SD-01) then problem investigators would follow all the downward-pointing arrows to try and trace the fault.   From IronMail-SD-01 I'd hop to EXCH-SD-05 and EXCH-SD-07, then possibly to the Mass Storage Device lower still. Although I may inform those using upstream CIs of my work, it's not where I'd go looking...

Hope that helps!