The bandwidth loading issue is due to the default behavior of ganglia
nodes, which is to have every node a listener as well as a broadcaster.
In effect, every node is listening to every other node, and attempting
to broadcast all its known information to the gmetad server.
We have cut down network traffic to a minimum by making only one node in
the data center the listener node (this also happens to be the gmetad
server and this is its only role in life), and all the cluster nodes
"deaf". This is a parameter in the /etc/gmond.conf file. That way, each
node in the cluster just sends out only its information and thus the
volume of traffic is very small in comparison. You can also tune how
often the data is sent.
On our high speed, low latency interconnect clusters (250+ nodes), we
are seeing very little ganglia traffic on the network... so small it is
unmeasurable. On the gmond listener/gmetad server, CPU load rarely tops 5%.
regatta wrote:
Guys,
We have a very old bad experience with ganglia (I think 2.5) and we
installed it in around 26 clusters with 128 nodes and as far as we
know it generate alot of traffic in the network and use alot of CPU
resource in the management node so we had to disable it since we need
the network bandwidth
Anyway today we have a new oracle cluster with around 48 nodes and we
want to ask you how do you think ganglia 3 ? does it produce alot of
network traffic, does it use the CPU alot ? did anyone use it with
oracle cluster ?
give me your feedback please
Thanks
