young wasp
We have a FlexGroup consisting of two volumes one on each aggregate (FAS2700) in a 4 node cluster. We use CIFS for the client protocol. As we copy many large files onto the FlexGroup we can see on the load on the two controllers holding the aggregates that with almost every file it switches from one to the other... which is expected... with "sysstat" we can see that when it writes to the node that has the LIF and to it's aggregate the speed is OK and the NICs are loaded with almost the same load as the disks are writing... just as expected. But when it has to write to the other node's aggregate if of cause have to relay the network traffic to the other node, so we see almost identical in and out loads on the NICs, but on the node that holds the aggregate it recieves the load on the NIC but reads as much as it writes... can anyone explain this ? I can understnad the writes, but why all the reads? The picture attached shows the load of that node... and just to be clear, there are no other loads on the systems while we upload the files...
marble nest
short wigeon