SnagAJob.com is a fast-growing, online job board that has no room for downtime. Every lost minute of uptime is potentially lost revenue. So a few years ago, SnagAJob.com realized it was time to switch out the inefficient hardware it had been using to manage its traffic and load balance its Web server farm. Around the same time, the Richmond, Va.-based online job site began expanding its computing environment into a 100 percent virtualized private cloud using blade servers running VMware.
Matt Reidy, director of IT operations at SnagAJob.com, says the company needed a product that could keep up with SnagAJob's growth. Launched in 2000, the site has been growing at double-digit rates year over year. At present it has more than 20 million members and counts McDonald's, Home Depot, 7-Eleven and Macy's among its customers. Just as important, he wanted something that meshed with IT's cloud-computing strategy. So when Reidy came across a software-based load-balancing, application and traffic management solution from Zeus, he was intrigued.
"We had some devices that were not working for us. We had an urgent need to switch to something that did work," Reidy says. "What drew me to the product was their software had the ability to run my load balancer as a virtual appliance in a virtualized environment. That was fantastic." Reidy says he downloaded the software to test it "and had something running in hours without even contacting anyone directly or talking with a reseller. So we tested the product to do Web farm load balancing and to load balance the Web traffic coming into our server farm. It worked great."
SnagAJob.com now has five instances of the application: two clusters of Zeus Traffic Manager, and a fifth running in SnagAJob's development and quality assurance environment. Thanks to the fault tolerant capabilities in Zeus' Traffic Manager, Reidy says SnagAJob.com is able to maintain 100 percent uptime because of redundancy created by the clusters. "My infrastructure and architecture design philosophy is to have everything redundant, whether physical or virtual. The ability to have clusters of Traffic Managers has been key so that if anything goes wrong we have a back up, and it is a seamless."
SnagAJob.com not only uses the Traffic Manager for its Web farm, but also to load balance and manage traffic going from Web farm down into the company's application stack, and to load balance read-only database transactions to a SQL server. The software has given SnagAJob.com insight, via activity graphs and real-time displays of traffic patterns, into its Web servers, application servers and database servers.
SnagAJob.com has been beta testing the newest version of Zeus's product, the Zeus Elastic Application Delivery platform, unveiled earlier this month. SnagAJob.com plans to deploy the new version sometime in the third quarter. The new version is designed to help organizations manage applications, servers and traffic that are spread across physical and virtual environments and technologies, as well as across different geographic regions. It's also aimed at helping organizations--many of which now have more tightly integrated networks and applications and which leverage a variety of end user devices, from mobile phones to laptops--to better understand how their applications are performing.
The Zeus Elastic Application Delivery platform now includes a Multi-Site Manager, which lets organizations tie together multiple, geographically dispersed clusters of Zeus Traffic Managers by providing a single interface to configure and manage them. The built-in Zeus TrafficScript lets businesses set and apply specific traffic management rules globally to accommodate changing business needs, such as prioritizing incoming traffic from key users.
SnagAJob.com's Reidy says the Multi-Site Manager capability is an important addition for his company, which currently operates only one data center but will likely add more as the company grows. An additional data center, perhaps on the West coast, would improve network latencies that naturally occur as users get further away from the company's Richmond data center.
Zeus also added a new auto-scaling feature that lets businesses automatically increase or decrease virtual and cloud capacity based on peaks in demand. Zeus' software will automatically deploy additional instances of an application if a pre-defined SLA, as set by the user, is exceeded. As service levels even out, the software can then switch off the added resource. Auto-scaling is initially available in Amazon AWS, VMware vCloud, Rackspace Cloud and RightScale-based environments. The upgraded platform also includes a new analytics module that provides real-time visibility into application performance.