On Tue, 30 Jul 2002 09:37:09 -0400, Bill Bitner <[EMAIL PROTECTED]> wrote:
>It depends. With Linux it depends even more. > >Remember that scaling involves both a software and a hardware MP factor. >On zSeries, we have some advantages from the hardware implementation. >There are some workloads that scale very well on Linux for zSeries. >Klaus Bergmann did a presentation at the last SHARE that gave some >examples of excellent scaling on a 16-way. Of course I'm sure that's >not the case for all workloads. :-) > >As for scaling with virtual MP support. There is a small bump/cost in >VM overhead in going from a virtual 1-way to a virtual n-way. However, >after that the overhead is basically linear up to the maximum of 64 >for a given virtual machine (though defining more virtual processors >than real processors will not be of any benefit and may just add >overhead). The ongoing work on NUMA support will eventually benefit the S/390 - z/VM Linux/390 environment. If multiple virtual machines need to work cooperatively, it is exactly like some NUMA machine where certain storage is under local control, some storage is shared (shared seqments) but need locking, and communication between machines has some special requirements. john
