Rai Expands Network Monitoring Tool

San Francisco-based data software vendor Rai Technology has added support for Thomson Reuters’ RRMP (Reuters Reliable Messaging Protocol) multicast data transport protocol to its Rai Insight multicast network monitoring platform, enabling clients to monitor unusual network activity and message re-requests that might indicate problems with RRMP-based Thomson Reuters Enterprise Platform (formerly known as RMDS) infrastructures.

Rai previously supported RRMP at the specific request of clients, and rolled out the new capability—which is already live at two global investment banks—into Insight earlier this month, in addition to the Thomson Reuters RRCP (Reuters Reliable Control Protocol), Tibco Rendezvous and Informatica Ultra Messaging and ciServer messaging layers already supported by the platform.

Insight provides real-time monitoring of software transport layers that sit atop multicast messaging layers in data infrastructures, and alerts users if a software application is excessively re-requesting data in a manner that could indicate a greater problem with the network or the application itself that might cause queuing and create “multicast storms” lasting from seconds to minutes that impact other applications’ ability to receive data in a timely manner.

“We can tell you where the problem is, what the problem is, and whether the problem is the infrastructure or actually something else—for example, a firm can avoid buying additional bandwidth if it understands that the problem is not a network issue, but is an isolated problem with one machine,” says Cameron Momtaz, chief operating officer of Rai. “So it is critical to understand why retransmissions are happening—for example, if one machine is down or just unable to keep up with the network.”

Networks Fit to ‘Burst’
Momtaz says a common cause of multicast network problems is running an application or spreadsheet that consumes a lot of data on an under-powered server that cannot keep pace with the flow of data, and is therefore constantly missing and re-requesting data.

“We’ve also seen cases of the network infrastructure itself hitting capacity constraints. If your data is very ‘bursty,’ it may look like you are only using 60 percent of your available bandwidth, but if you look at the bursts, you might be hitting capacity, which can be very risky,” he says. “The worst-case scenario is that you make a loss because data is not getting through and your applications aren’t seeing the data they need—though in most cases, the data session layer will tell you something’s wrong, so you won’t be trading on bad data.”

The system uses “collector” devices that act as network taps and join multicast groups to monitor traffic, then displays the data in a user interface for analyzing historical activity and alerting users—such as network administrators, market data professionals, or even developers who want to test the traffic profile of their applications—if retransmissions exceed a certain amount or percentage of bandwidth capacity.