 |
How We Tested
Our existing production environment hosts more than 600 user nodes, with a mix of Linux, Microsoft Windows NT and Apple Macintosh OS X servers on the back end. All users have Internet access. Multiple externally accessible Web sites are hosted on Red Hat Linux and Mac OS X-based Apache servers. Our normal primary and secondary DNS servers are hosted on Red Hat Linux, running BIND 9.2 that we manage with a GUI (QuickDNS 4.x from Men&Mice). Our primary DNS resides behind our firewall; our secondary DNS lies outside.
Each appliance was installed behind our firewall one at a time over a six-week period. Our existing DNS configuration was imported, and the appliance was activated as primary. Each appliance then served two weeks in production as our domain's primary DNS. Each product took its turn as our main internal DNS server. During that period, the switch to any one of the three appliances was transparent to our users. Real DNS lookup times appeared anecdotally similar to, if not faster than, our end-user population. We performed multiple stress tests using the queryperf tool from ISC, while the appliances and Linux box were connected via a private 100-Mbps connection. The appliances were configured as primary and were not managing any production queries during stress testing.
|
Discover the benefits of scalable backbone and network services for application and cloud providers.
Network automation is gaining momentum. Here's how you can drive this powerful technology to its full potential.
Subsea cable alternatives can provide a level of comfort for those concerned about disruptions. Realistically, they will continue to play a small, but critical role in the short term.
