pros and cons relative to what?
If you are interested in recording, then this is covered in the
"
Developing Client-side Call Recording Applications using Application Enablement Services "
https://www.devconnectprogram.com/fileMedia/download/7eeb6799-15a6-4701-ae96-57d819b482eb
A basic DMCC license does not include station or soft phone licensing. This type of licensing is mostly used by customers who have an overabundance of these forms of licenses. So depending on the specific customer the need for basic/full DMCC licensing will vary; for the most part a full license should be assumed for both MR and SO recording.
DSP resources are allocated when the device is active in a call. I believe that even in the state where the SO party is off hook idle waiting for the target to become active on a call, the DSP is in a reserved state. They are not reserved otherwise.
I am not sure what 'parallel' recording is in your terminology, I probably would have said redundant, duplicated or high availability. In the past SO participants could be configured to receive unique media streams from the call. MR participants could not be configured in a similar way. In AE Services/Communication Manager 8.0.1 there are significant enhancements related to split stream (aka stereo) recording. Up to 6 MR devices are now supported, and can be used to record 6 specific media streams (including all parties) from a single call. Please see the DMCC programmer's guide for additional information.
SO can support two SO participants in a single call when Communication Manager is configured properly. This was introduced in Communication Manager Release 4 or 5 I forget which. Note that SO reduces the total number of parties allowed in a call (total limit is 6). SO can be used in silent or active mode. Silent SO participant uses the same number of time slot & DSP resources as MR, whereas active mode uses an additional times lot and half of a DSP resource to support the SO party talking into the call.
By its nature MR is not obvious to call participants. When active SO is used, call participants are aware there is an additional party in the call. When silent SO is used call participants are not aware there is an additional party in the call (unless the party limit is 'unexpectedly' hit).
SIP endpoints can be involved in calls where SO is used (i.e. the target of service observing).