I have fought with the definition of "End-To-End" monitoring within our shop for several years. While CEM/APM provides quite a bit of coverage, the lack of detailed out of the box system monitoring (infrastructure) is far lacking.
We have implemented quite a few of the field epagent Perl scripts to provide a bit more coverage and we are looking at Unified Infrastructure Management (UIM formally NimSoft) to try to provide more DB2, MS SQL Server and infrastructure monitoring.
With the APM, the coverage will depend on your implementation since there are several SOA interfaces that are not covered, such as IIOP, or ESB using IBM IIB (broker) communicating with CICS on IBM mainframe/COBOL.
CEM is only really helpful if the business and application development chip in with the definition of the business services and business transactions. In my experience without direct involvement from business and application development, using CEM is very difficult to form actionable and realistic information.
With the concept of End-To-End monitoring there is also the thoughts of real time (real user experience monitoring RUE) and also synthetic monitoring (typically called the 3 AM monitoring).
The CEM/APM solution does rather well with real-time monitoring from network entry to application host (J2EE application server) but without the next jump from application server to server/database, doing root cause becomes hunt and peck. The browser real time monitoring solution might give you from the end user browser to your network.
For the 3 AM solution, the CA solution is the Cloud monitor service, but if you are in a shop that will not permit use of cloud services, then you will need a different solution/software. We are using HP Sitescope to provide some coverage to our entry URLs with a SiteScope server being hosted on a DSL line, but that solution has it's issues also.
With the end-to-end monitoring, there is quite a bit of department monitoring solutions (Splunk, VEEM, NetQos, Qpasa, various IBM tools, home-brew scripts, What's up Gold, Google Tools, Urchin, Run of the mill Windows server tools) that will also cause the old argument "But X says this and CEM/APM/UIM does not, so I'm not going to look at CEM/APM/UIM"
So, if you are going down the End-to-End monitoring path, get solid requirements(definition), buy-in and support at the highest point within your company and backing from all the numerous layers that has responsibility to provide quality service/application/data to the customer/end user/people who give you money to do a job.
Once you have backing, and you decided to use CA software, have a standing order with CA Professional services to get feet on the ground to support, customize, configure, trouble-shoot, implement, design, train users. The CA APM/CEM/UIM and various other CA products are complex and the complexity will grow depending on your implementation. Make sure that the solution/business/application/service architects at your company not only supports your End-To-End monitoring effort but are 100% engaged with their past, current and future application/service designs so that your choice in End-To-End monitoring can support and monitor their decisions.
Please do not fool yourself thinking that a few people will be able to install, configure, train, help do root cause analysis after a few CA APM/CEM courses. Doing so will reduce your return-on-investment and your adoption rate will be slow and painful and depending on your total support for End-To-End monitoring may end up with a gutted budget, then abandonment.
Do hope this helps,
Billy