We recently had a DDos attack on our network that flooded our network with UDP traffic taking our network and internet connections to a crawl. After this the effects were that random clients suddenly took up to an hour to login. We have 2 Cisco Cat 6509s as the core switches and Cat Cisco 400x, 2800, and 350x servicing our user base and there is no certain switch that it doesnt happen on. The client will login and it will load their profile for an hour. We changed the registry in the Kerberos key to use TCPrather than UDP for packets with a sizeover 1K (esentially all Kerberos packets) and now the clients work. We are rebooting are 6500s this weekend to clear out any possible "garrbage" that may have resulted from the attack. Has anyone seen this issue before???? Everything worked before the DDos so portfast trunking etc etc all were fine and worked and everything is still the same.
