• Data centers hosting desktop virtualization applications like Citrix and Web 2.0 services often exhibit poor performance if the network exhibits uncontrolled or intermittently high levels of latency.
• Collaborative applications like Cisco TelePresence require consistent delivery of controlled latency and low jitter in the network.
• Algorithmic trading and distributed grid computing require ultralow levels of network latency and loss (typically less than 5 milliseconds and no packet loss).
• Unprecedented visibility of network events that affect application latency and loss increases confidence in the network's ability to handle latency-sensitive applications.
Cisco BQM microvisibility is the ability to detect, measure, and analyze latency, jitter, and loss affecting traffic events down to microsecond levels of granularity with per packet resolution. This helps enable Cisco BQM to detect and determine the impact of traffic events on network latency, jitter, and loss. It is critical to measure at this level of granularity as a traffic microburst on a data center Gigabit Ethernet LAN, lasting only tens of milliseconds, can easily cause hundreds of milliseconds of queuing latency and thousands of lost packets on the WAN.
• Rapid troubleshooting of network congestion issues reduces the time, effort, and uncertainty involved in resolving issues.
Cisco BQM detects network quality performance (congestion) violations against user-specified thresholds in real time on all interfaces. Simultaneously Cisco BQM operates a background rolling packet capture. Whenever a threshold violation or other potential performance degradation event occurs, it triggers Cisco BQM to store the packet capture to disk for later analysis. This allows the user to examine in full detail both the application traffic that was affected by performance degradation ("the victims") and the traffic that caused the performance degradation ("the culprits"). This victim/culprit analysis is vital to understanding what is really happening during performance outages and what needs to be done about it.
• Understanding how to provision the network for the desired level of latency and loss cuts the cost and time to support latency-sensitive applications on the network.
Cisco BQM monitors per packet network performance against user-specified quality targets for delay and loss to provide definitive early indication of network performance degradation. Because it operates at the network layer, Cisco BQM is then able to provide detailed bandwidth and quality of service (QoS) policy provisioning recommendations, which the user can directly apply to achieve desired network performance.
Cisco BQM Functions
• Network Service Quality: Cisco BQM measures the loss, latency, and jitter experienced by the actual application traffic between any two BQM appliances in the network. By passively observing the packets at two or more locations and sending "out-of-band" timing information to a designated BQM appliance, extremely precise information on how long each packet was delayed (and whether it arrived at all) is presented to the user.
– Microsecond end-to-end measurement of latency, jitter, and loss of application packets
– One-way measurement in both directions
– Passive - does not affect the application traffic
– Supports all IP traffic - UDP, TCP, multicast
– Layer 7 Deep Packet Inspection (DPI) and automated application discovery and recognition
– Optional GPS time synchronization interface supported
– Programmable 1 in n packet sampling (n = 1 - 1000) per class
– Full multiple QoS class support
– Zero configuration for remote appliances
– Reroute detection for support of dynamic routing, for example, MPLS VPNs
• Expected Quality: Cisco BQM calculates in real time the level of loss and queuing delay expected to be induced by observed traffic as it encounters downstream low speed (bottleneck) links or QoS shaping in the network. It allows the user to understand the amount of performance degradation due to traffic-induced congestion versus performance degradation in the network cloud. Traffic induced congestion can be eliminated by adding the appropriate amount of bandwidth to the identified bottleneck point in the network. If the Expected Quality is a small contributor to the overall end-to-end network service quality, then the addition of local bandwidth is unlikely to help. Expected Quality is also used to estimate the network service quality characteristics from a "one-sided" deployment, that is, in situations where it was not possible or desirable to place a BQM appliance at the remote end.
Expected Quality is computed per packet, and distributions for packet delay and loss are available every five minutes. Supported QoS mechanisms in Cisco BQM 4.0 include:
– Class-Based Weighted Fair Queuing (CBWFQ)
– Low-latency queuing (including traffic policing)
– Link efficiency mechanisms (LFI, cRTP)
• Event Analysis: Cisco BQM allows users to set thresholds for network service quality (loss, latency, jitter), Expected Quality (delay and loss), Corvil Bandwidth (delay protected, loss protected), and Traffic Insight (microburst rate, packet rate, and so on) measurements. These thresholds define bounds of normal or expected levels of performance on each managed connection. If any threshold is violated, Cisco BQM triggers a 10-second leading and trailing trace capture around the event, and a quality violation alert is raised (optionally causing e-mail notification or a Simple Network Management Protocol [SNMP] trap). The event traces record every packet on a managed interface including:
– Packet-size and hardware-resolution time stamp
– IP source and destination addresses and ports
– Recognized application
– Measured end-to-end delay
– Expected bottleneck delay, and whether dropped or policed
– Corvil Bandwidth (bandwidth required to protect the packet)
Cisco BQM provides postevent analysis capabilities at the site, class, application, conversation, or even packet level. The GUI presents all statistics measured during the event including:
– Time series and distributions
– Top-N tables (talkers, listeners, applications, and flows)
The user can zoom into any subinterval of interest during an event down to submicrosecond resolution, supporting:
– Fine timescale inspection of critical phases of events
– Detailed "victims" and "culprits" analysis to identify key traffic sources and sinks during events
• Bandwidth Sizing: Cisco BQM uses a technology called Corvil Bandwidth to determine the bandwidth required by application traffic to be protected from exceeding user specified quality thresholds. Bandwidth sizing policies for classes or interfaces are expressed using a:
– Threshold for queue delay (1 millisecond to 1 second) and queue size (1 packet to 2000 packets)
– Percentage of packets to be protected (1 percent to 100 percent, in 0.0001 percent increments)
– Period for which protection policy applies (5 minutes; 1, 2, or 4 hours; 1 day; or 1 week)
– Delay Protected Corvil Bandwidth - the bandwidth required to protect traffic from experiencing more than 50 ms of queuing delay for 99.99 percent of packets every 5 minutes.
– Loss Protected Corvil Bandwidth - the bandwidth required to protect 99 percent of packets from exceeding a queue limit of 128 packets every hour.
Cisco BQM computes Corvil Bandwidth in real time per class or per interface (FIFO) and uses the results to make specific bandwidth and QoS policy recommendations to deliver the required levels of network performance. Specific recommendations include:
– Link bandwidth sizing
– Class bandwidth sizing
– Policer settings
– Queue limit settings
• Traffic insight: Cisco BQM contains a Layer 7 Deep Packet Inspection (DPI) engine that operates at microsecond granularity. This provides unprecedented traffic insight and gives a detailed real-time view of how network resources are being used to support the network application environment. Cisco BQM traffic insight includes the following:
– Autodiscovery of network applications with Layer 7 signatures - approximately 350 signatures built-in, and can be easily expanded
– Microburst detection of programmable duration from 1 millisecond to 1 second with rate shape analysis
– Top talkers, listeners, and conversations
– Reports average link use, packet rate, and packet size distribution
• Quality Alarms: Cisco BQM provides for detection and analysis of various events to external systems through fully configurable SNMP traps, syslog streaming, and e-mail alerts. Cisco BQM alarms can be based on violations of the following thresholds:
– Network Service Quality (end to end loss, latency, jitter)
– Expected Quality (delay and loss)
– Corvil Bandwidth (delay protected, loss protected)
Table 1. Features and Benefits of Cisco Bandwidth Quality Manager 4.0
Table 2. Hardware Specifications
Service and Support
For More Information