Hi Folks,
L4Linux.org mentioned: "Compared to monolithic Linux, there is a small performance tradeoff because of the µ-kernel architecture. However, the initial L4Linux has been somewhat optimized, and on L4/x86 it has a very acceptable slowdown of less than 4 % for any relevant load. " Any place can find more detail about the test?
And I'm very curious the L4Linux performance vs "using hardware virtualization on top of L4", have you made any comparison before?
-- Thank you! Bob
Hello
L4Linux.org mentioned: "Compared to monolithic Linux, there is a small performance tradeoff because of the µ-kernel architecture. However, the initial L4Linux has been somewhat optimized, and on L4/x86 it has a very acceptable slowdown of less than 4 % for any relevant load. " Any place can find more detail about the test? And I'm very curious the L4Linux performance vs "using hardware virtualization on top of L4", have you made any comparison before?
It depends. From my experience, I would say, a computation-intenive application demonstrates the same performance as a native application. I/O intensive is another case, here the difference can be from -something (well, uKernels can be faster sometimes) up to +unlimit. Several years ago we measured a network performance from one L4Linux instance to another. At the beggining the performance was near the 400Mb, which was comperable with performance of two VBox-based virtual machines.
On Wed Feb 07, 2018 at 08:02:15 +0000, Bob Liu wrote:
L4Linux.org mentioned: "Compared to monolithic Linux, there is a small performance tradeoff because of the µ-kernel architecture. However, the initial L4Linux has been somewhat optimized, and on L4/x86 it has a very acceptable slowdown of less than 4 % for any relevant load. " Any place can find more detail about the test?
And I'm very curious the L4Linux performance vs "using hardware virtualization on top of L4", have you made any comparison before?
To add what Vasily already said, L4Linux has a decent performance but cannot cope with what hardware features can achieve, esp. for the memory and CPU virtualization. For I/O there isn't really an architectural difference, except again if there's hardware support, such as interrupt injection. For L4Linux there's the (famous) paper from 1997 describing it in its first variant: https://os.inf.tu-dresden.de/papers_ps/sosp97.pdf
Adam
l4-hackers@os.inf.tu-dresden.de