By Antti Pennanen -
Hey everybody,
Our client is running a HP Blade and VmWare based environment. The interconnects between blades are 1gbit (6 interconnects in total). For the network traffic there are 2 switches in an active-active HA configuration. The uplink is 100mbit/s and all the traffic is routed through VRRP to a StoneGate firewall which then does nat or forwards the traffic to the correct VLAN.
Today our monitoring noticed that some of the services were down so we started to investigate the problem. We soon noticed that the uplink is basically completely saturated and that is the reason for the weird network behaviour. Later we found out that the root cause was a malware on one of the virtual machines which was sending huge amount of DNS traffic out. (350mbit/s, which obviously couldn't get through)
I started thinking that is there any best practises to avoid this kind of situations? So that one VM (or any network client) couldn't completely saturate the uplink. Anyone here experience of this kind of network problems? Flow-Control? QoS?
Any ideas are welcome and thanks in advance!
Andy
To unsubscribe from the list send a blank e-mail to mailto:studiosysadmins-discuss-request@studiosysadmins.com?subject=unsubscribe
↧