improvements we see in Oracle Linux not ported to RHEL
Why Red Hat does not include the extra Linux special features that Oracle have given its customers:
such as Dtrace, and Ksplice, not to mention the ability to update to the latest upstream kernels.
Responses
Oracle specifically bought (and subsequently made an Oracle-exclusive) the IP from the Ksplice project. So, the simple answer would be "licensing".
The dtrace utility is part of the spoils of Oracle's acquisition of Sun. dtrace is only ported to UEK because Oracle wants to keep the utility as a value-differentiator between UEK and any other Linux releases. While there are ports of dtrace for Linux, it's still kind of a work-in-progress for that platform. I imagine that the use of CDDL for the code - and that any kernel running it would have to run it as "tainted" code - doesn't help with RedHat wanting to put it into the core release. If dtrace is critical to you, you can always pull it from GitHub (not currently seeing it in EPEL).
Also, you can always update to the latest upstream kernels ...you just lose most (if not all) of the protection your support contract gets you. The basis of commercial releases is that they have contract-support available to them. Such support is maintainable only by having a standard set of binaries all tested to work together. Asking why RedHat doesn't foster use of upstream kernels in production code is like asking why Oracle/Sun didn't provide contract support for OpenSolaris.
Would pulling dtrace from GitHub involve adding a repository with yum?
No.
Think of GITHub directories as a mini repository with history and version tracking ( and branches) for a software package.
I believe Dtrace leverages SystemTap too. As a Solaris user and big fan of Dtrace. (IMHO) It is too much of a hack in centos/rhel. Good for education purposes but not useful in real world.
Github is for pulling what is, essentially, "raw code". Much of what's stored there is source code and not stored in a yum-compatible format.
You'd need to use a git client to pull down the software, then compile and install it yourself. You'd also need to manually do requisite dependency tracking to find and compile/install all of the other things that the dtrace port relies on.
Welcome! Check out the Getting Started with Red Hat page for quick tours and guides for common tasks.
