I see documented that nameloc runs best with 2-3 instances on a network.
What does best mean and what issues can I expect if I have more than this?
Do these issues increase with the system being busier.
Thanks.
Doug Rixmann <rixmannd@rdsdata.com> wrote:
I see documented that nameloc runs best with 2-3 instances on a network.
What does best mean and what issues can I expect if I have more than this?
Do these issues increase with the system being busier.
There are a couple issues – one is network traffic. Each nameloc
will spend some time scanning the network, checking for up/down state
of nodes in order to track ownership of names and stuff like that –
the more of them you have, the more network traffic overhead you will
have. Also, the more of them you have, the more likely you will have
two different namelocs with inconsistent (not yet synchronised) views
of the current state of global names on the network. If you go beyond
10, you are almost guaranteed to have inconsistent global names, because
the itnernal lists of “other name locator nodes” used in various places
maxes out at 10.
2 to 3 is a number suggested as a reasonable compromise between
redundancy (single node failure won’t prevent lookups) and
consistency + overhead. If you KNOW a particular node will ALWAYS
be up and connected, just running 1 nameloc on that node works best.
-David
QNX Training Services
http://www.qnx.com/support/training/
Please followup in this newsgroup if you have further questions.