I would like to see a CHR ISO image.
I would like the ability to directly install a CHR ISO onto a server (skipping & not using a hypervisor). Thus a CHR could have direct access to all hardware devices and eliminate the need for a hypervisor.
How is this different from the normal x86 install that has literally been an option for over a decade? Download:
https://mikrotik.com/download (X86 section, right after ARM - the CD image is literally an ISO as you requested) and license info:
https://wiki.mikrotik.com/wiki/Manual:L ... nse_Levels (Non-CHR X86 licenses are feature-based, not port-speed-based). CHR was introduced because the basic X86 version assumed it was running on bare metal, and wasn't properly suited for being installed as a VM.
Fist - x86 is a 32-bit system
Second - x86 does not have any paravirtual support
Third - From my hand-on real-world experience using x86 as the 207.32.195.2 public btest server, I would experience a full x86 lockup sometimes 4 times a day. Now that I am using a CHR 64-bit system with some paravirtual device support, it has NEVER locked up.
Fourth - a 64-bit system directly supports much much larger access to big memory.
Fifth - I have no idea if a x86 32-bit ROS system could even hold a full ipv4 & ipv6 bgp routing table without problems - let alone run at 5-gig plus sustained speed without locking up many times a day.
Sixth - 64-Bit operating systems also support 64-bit access to memory (faster)
Seventh - 64-Bit systems also can also use more efficient 64-bit compilers and use 64-bit instructions (faster again)
Eighth - I believe 64 bit systems also have additional registers (more registers than 32-bit systems) (again faster again)
Ninth - a fully 64-bit compiled system can process more instructions (some new 64-bit instructions) which can result in fewer CPU clock cycles need to perform a function. (((64-bit hardware can easily out-run 32-bit hardware with both CPUs clocked at the same CPU clock speed.
Tenth - newer hardware devices (40-gig network cards for example) usually come with **ux optimum 64-bit drivers. (again - again faster again).
Eleventh - a 32-bit system running 32-bit code can use twice the CPU/memory read/write clock cycles as a 64-bit system running 64-bit code. (again - lots of these - faster again)
So I ask - what good is a system (possibly 32-bit) on a multi-many core system if the clock speeds of the CPU cores is sllooowww and has little or nooo built-in CPU cache. In 5 years, we could have 10-Ghz Xeon CPUs 25-+ core with multi-gig CPU cache per core. It will happen sooner than you think. My first computer was a 4-bit PPS04 running at 300 Hz (then upgraded to the 8008 - then the 8086 and now the 64-Bit Xeon CPUs - my how things get faster.
The first hyper-visor I ever experienced back in the late 80s/early-90s took 3 days to boot DOS. So , even though hyper-visors are faster - a no-hyper-visor system is the fastest least overhead method to make something run incredibly fast. Thus a CHR ISO could kick start a 40-gig L3 full-network-speed routing platform.
North Idaho Tom Jones