Sometimes, despite all of the devices on a network, the infrastructure can seem invisible to IT and network managers.
That’s especially true thanks to virtualization and load balancing application delivery controllers, which hide what’s going on in parts of the network and make it hard to collect data for analysis.Riverbed
Technology Inc. says the latest release of its Cascade network performance management software gives deeper visibility into the data centre.
Version 9.5 – to be released this quarter – includes the ability to monitor across application delivery controllers, and the ability to see into VMware’s ESX virtual environments.
It’s not that managers don’t have tools, said Heidi Gabrielson, Riverbed’s senior product marketing manager, but they are “living in the dark ages” using SNMP-based fault-monitoring or packet-by-packet trouble-shooting.
With the advent of load balancing, virtualization and cloud computing, “these new technologies are causing 21st century problems, but IT operations is using 1950s technology,” she said.
To get into the virtual world, the new version will offer Virtual Cascade Shark, which sits in ESX environments and monitors the virtual switch. It sends metrics to Cascade Profiler, the application’s reporting and analysis engine.
As an example of the potential, Gabrielson said a beta customer that was suffering 50 “events” a week with its virtual server cut the occurances down to one a week after Virtual Shark discovered the problem came from a service provider’s firewall.
To give application delivery controller visibility, Cascade will collect data from ADCs such as F5 Networks Local Traffic Manager as well as Riverbed’s Stingray Traffic Manager.
The additions impressed industry analysts.
“What Riverbed is doing is tying the physical and virtual worlds of (network) information together in a way few companies are doing,” said Andre Kindness
, enterprise networking analyst at Forrester Research.
In particular, he called Virtual Cascade Shark “cutting edge.”
The improved ability to grab packet and other data is vital to achieve the goal of automating infrastructure management, he added. “You’ve got to monitor before you can react.”
Zeus Kerravala, principal analyst at ZK Research, said the ability to see ADC controllers is “a pretty cool feature,” one that few other management tools have. Generally, the controllers “make network management systems blind,” he said.
“Cascade had been playing a little bit of catch-up in that they were trying to play in the performance management space with companies like NetScout and OpNet, but they didn’t quite have the number of features.” Not so now, he said.
Other new features in Cascade 9.5 include
--Multi-segment analysis, the ability to analyze related traffic streams captured from multiple locations or sources;
--Precision time-stamping support within Shark for nanosecond time stamps from SPAN port aggregators such as Gigamon GigaVUE, cPacket Networks’ CVU and cTap, and VSS Monitoring Distributed Traffic Capture. Time-stamped data is valuable in troubleshooting time sensitive applications such as financial trading environments;
--Voice over IP quality reporting by supporting VoIP protocols and quality metrics into Cascade Profiler.
Riverbed [Nasadq: RVBD] Cascade 9.5 is a free upgrade for those on existing support contracts. However, Virtual Cascade Shark is priced separately and depends on the number of hypervisors. Prices start at US$1,200 for up to four hypervisors and up to 2 TB of storage.
Cascade appliances start at US$35,000 for Cascade Express, which supports 25,000 Kb of data flows per minute. A high-end solution, which costs US$165,000 and supports 200,000 Kb of data flows a minute, includes the Cascade appliance, a gateway flow collector and two Shark packet capture appliances.