For a OS developer beginner, as I am, which design approach is better to follow:
- The microkernel approach design
- The monolithic approach design
Thank you
pepito
Microkernel vs Monolithic design
-
VE3MTM
RE:Microkernel vs Monolithic design
This is really the ultimate question of choice in the design of an OS. There are advantages and disadvantages to both.
Monolithic kernels have an immediate advantage of performance. They call functions in other parts of the kernel like in normal programs. However, failures in one part of the kernel is most often fatal. There is no way (in general) to restart a module if it has a problem. They are easier to write, because there is no messaging code for interprocess communication.
Microkernels tend to be slower (by how much depends on the implementation), but more stable. Just like with normal programs in an operating system, parts of a microkernel can be restarted if they crash. Extra support code is required to allow for clean, efficient interprocess communication and control. Because of this, it takes more work to get a microkernel up and running
Really the choice comes down to you. If you want a nice, clean, expandable system, you might want to op for a microkernel, otherwise you might want to write a monolithic kernel
Monolithic kernels have an immediate advantage of performance. They call functions in other parts of the kernel like in normal programs. However, failures in one part of the kernel is most often fatal. There is no way (in general) to restart a module if it has a problem. They are easier to write, because there is no messaging code for interprocess communication.
Microkernels tend to be slower (by how much depends on the implementation), but more stable. Just like with normal programs in an operating system, parts of a microkernel can be restarted if they crash. Extra support code is required to allow for clean, efficient interprocess communication and control. Because of this, it takes more work to get a microkernel up and running
Really the choice comes down to you. If you want a nice, clean, expandable system, you might want to op for a microkernel, otherwise you might want to write a monolithic kernel
-
pepito
RE:Microkernel vs Monolithic design
Thank you very much!
If the monolithic kernel is easier to write, then I prefer it to begin with.
Just a little question: What is the meaning of interprocess communication?
pepito
If the monolithic kernel is easier to write, then I prefer it to begin with.
Just a little question: What is the meaning of interprocess communication?
pepito
-
carbonBased
RE:Microkernel vs Monolithic design
IPC, in general, is a method for two programs/processes to talk to each other.
Typically, each process in an OS is in its own memory space, which means it has no knowledge, nor access, to any other process. IPC allows these two separate processes to communicate to each other.
You can implement it any number of ways... with a mailbox metaphore, using interrupts, with another "global" IPC process, etc, etc. There're many ways.
Cheers,
Jeff
Typically, each process in an OS is in its own memory space, which means it has no knowledge, nor access, to any other process. IPC allows these two separate processes to communicate to each other.
You can implement it any number of ways... with a mailbox metaphore, using interrupts, with another "global" IPC process, etc, etc. There're many ways.
Cheers,
Jeff
-
pepito
RE:Microkernel vs Monolithic design
Sorry,
What kind of information is exchanged between process?
Why is important the processes to communication?
pepito
What kind of information is exchanged between process?
Why is important the processes to communication?
pepito
-
VE3MTM
RE:Microkernel vs Monolithic design
In monolithic designs, processes (essentially programs) don't communicate between each other too much, unless you consider network sockets to be IPC. However, in a microkernel, this is essential. Say, for example, you have a working microkernel. You'll have things like file system drivers running as seperate processes. If a program wants to write something to disk, it will query the file system process to write the data to the disk, and in that query will be the file name, the data, maybe attributes for the file.