Oracle Consolidates Network Monitoring, Management Onto Single System

With the $7 billion acquisition of Sun Microsystems now completed, Oracle has to watch over more than 13,000 network devices and services--everything from access switches, core network routers, wide-area-network (WAN) routers, firewalls, load balancers and more--and it takes a comprehensive network management system. But what used to be an unwieldy patchwork of various network monitoring and management tools, some purchased, some open-source and some home-grown, has been whittled down to one: an IT management system from Monolith Software.

"We had numerous tools," says Craig Yappert, Oracle's senior director of Enterprise Monitoring Solutions. "We couldn't sustain that model." What Oracle wanted, and began looking for more than a year ago, was technology that would let the vendor compress and consolidate the various products and solutions it was using to monitor its vast corporate network, which stretches across continents, support its more than 85,000 employees, and is used not only to support Oracle's business applications but also customer service solutions, collaboration suites and even product development. The network also supports millions of customers that use Oracle on Demand, an outsourcing solution that provides companies with hosted Oracle applications such as customer relationship management (CRM).

The right technology would have to scale to the size and scope that Oracle was to become (at the time it was just beginning the process of buying out Sun) and beyond. "The challenge was not so much the feature and functionality of a solution, although that certainly is important. It was scaleability and adaptability," Yappert says. "It needed to allow us to consolidate all the tools we had, it needed to scale to the size and scope of Oracle today and in the next few years, and it needed to allow us to get at the data and repurpose the information the way we needed to."

On that last point, Yappert says Oracle wanted to be able to turn the network monitoring and management data such a system collects and present it to the Oracle end users as useful information about the network's performance. Oracle found Monolith Software almost by accident. It was reviewing different vendors' network management solutions and only considering using a product from Monolith Software for syslog management. But after meeting with the company, which was founded in 2005 by Jeff Parker and Shawn Ennis, Oracle realized that Monolith Software had much more than syslog management.

Monolith Software was started specifically to help organizations rein in their network monitoring tools with a modular, Web-based solution set, and Parker and Ennis drew on their years of experience as consultants helping organizations monitor networks. "When companies bought software for the network operations center, traditionally they had to by from several different providers, and then hire a consultant to put it all together and train the staff," Parker says. Consultants would "take all the piece parts and crudely stitch them together, but to maintain that Frankenstein, they needed a lot of manpower. So we built our own software from the ground up to address these key issues."Oracle began implementing Monolith Software's solution a year ago for fault, availability and performance monitoring. The solution polls all the network devices and services in a hierarchical, leveled fashion. Some devices are only pinged for faults using syslog and SNMP traps. Other, more critical devices are monitored more thoroughly using deep SNMP polling mechanisms for performance, traffic usage and CPU memory analysis. All the data culled from the devices is sent to a central repository where a rules engine analyzes the data and generates, if necessary, work tickets to Oracle's network support team. On average, about 4,000 tickets are generated each month, Yappert says.

Because the tickets are integrated so closely with the repository and rules engine, any changes to network events are automatically reflected in the work tickets. "Some of the issues go away on their own. If the event clears itself, Monolith automatically clears it too. So the system is not just spewing noise, but helps with the flow and management of the work," Yappert says. The network data is also merged with all the information Oracle collects during network discovery, as well as with any fixed asset information (such as the cost of an asset, when it was purchased, when it was put into service, where it is located, and even what service or organization depends on that device). "All that information is provided to the network support analyst as they respond to tickets, so they can consider whether it is time to replace a device, or focus on another ticket because the device is not servicing a critical application," Yappert explains.

One of the highlights of the Monolith Software solution is the dashboard engine, which creates Web-based, real-time views for line of business managers and others that illustrate network performance as it relates to Oracle's business. The dashboards can show capacity and performance for latency, bandwidth, voice quality scores, events, and even the availability of all links and devices within any specified line of business or Oracle region. Since implementing Monolith Software's solution, Oracle has been able to clean up device configurations, thanks to the cohesive view of network monitoring. It has also been able to track how and where specific network events impact other devices.

Oracle has also reduced the overall number of servers supporting its network monitoring and management processes by about 50 percent, which in turn is reducing its power and rack footprint. The company, which went live with the Monolith Software solution across all its network operations, including Sun's, at the end of February, is also already starting to see a reduction in the time it takes to resolve issues. "We are just starting to measure that data. How that reduction translates into how many people we have out in the field or taking support questions is unclear. But we do think we will be able to solve problems faster," Yappert says.

Before the full implementation, Oracle completed a in-depth business case that included a return on investment (ROI) analysis. The ROI analysis showed that Oracle could save $200,000 in the first year, mainly because it wouldn't need to expand its team of IT managers that oversee the network and because it would save on licensing costs and hardware costs. "I have to say I was skeptical in the beginning, but it has really proved itself out," Yappert says.