During this past week, the university community experienced intermittent outages and performance issues with network connectivity. On Monday, August 31st, the main core DNS server displayed high CPU utilization and began to exhibit slower than normal responses. This had a cascade effect on other systems and services, such university authentication. Our main wireless controllers themselves also exhibited issues which resulted in clients not reliably associating with local access points. We have a replacement DNS and DHCP infrastructure that is scheduled to be deployed this semester, but in the interim, we have addressed performance issues by separating DHCP services from the main campus DNS server. The issue with our main wireless controllers is more complex. While the load is not disparate with respect to last year, it has become increasingly asymmetric. However, while the current usage was within the capacity parameters of the controllers themselves, one or more of them stopped providing service. We have opened a support case with our vendor and rebalanced load across the infrastructure to decrease the chance for outages. We have also begun redesigning our wireless architecture based on their feedback and the performance characteristics that we have observed.
Network connectivity is a critical, highly visible, and widely used service, and we understand that any outage has a significant impact on our community. We will keep you informed as we identify the underlying problems and as we take additional actions to remediate. Please do not hesitate to contact me if you have comments, questions, or concerns.