[erlang-questions] Erlang R13B Multicore Efficiency Questions

Greg Perry <>
Tue Jun 2 21:17:13 CEST 2009


Part of the issue is long term scalability, being able to build a
prototype system that is decoupled from the host hardware and which can
be seamlessly migrated to a larger hardware cluster is important.  I
haven't quite figured out the limitations of Mnesia as an RDBMS, ie if
it will scale to many millions of records without resorting to some form
of distributed keystore or database sharding process.  Being able to
build the entire system virtually and then migrate those VMs to a larger
hardware cluster as the workload increases is of benefit, in addition to
some of the automated workload balancing and high availability options
inherent to VMware ESX/ESXi.

The other half of the question relates to virtualized cloud computing
infrastructure, for example deploying Erlang on Amazon EC2.  VM instance
selection with Amazon AWS varies in price depending on the processor
count you select (non-SMP VMs on their cloud go for ~$70 per month; SMP
"enterprise" instances are priced several orders of magnitude higher and
you pay a premium for multiple proc/core VMs on their backbone).

>I don't understand why you need to involve the VMware stuff and
virtualization at all if all you want it to build a highly parallell
application with high performance.
>Why not just run 1 OS instance (Linux would do well) over the whole
multi-pro and multi-core system.
>On top of that you run 1 Erlang VM with as many internal schedulers as
there are logical processors (i.e. cores and hyperthreads).
>
>You will then get a very simple systems that scales very well.
>
>The Erlang VM with SMP support is very performant and reliable. It is
used in many "non-stop" running applications.


More information about the erlang-questions mailing list