Palermo Performance Analyzer Concepts
Some of the Palermo Performance Analyzer metrics are unique. Here is quick walkthrough the concepts and related metrics available in the Analyzer.
Limitation Analysis for Throughput
Analyzer classifies each flow according to its own limitation for rate increase:
● Network Limitation
● Data Generation Limitation
● Flow Control Limitation
Aggregating traffic visualization on each of these categories,
allows network managers to get useful insight on network capacity and traffic profile.
Network Limited (red) profile for typical Content Delivery Network downstream traffic.
Data Generation Limited (green) profile for typical Content Delivery Network upstream traffic.
Throughput for Network Limited Flows
Measuring throughput of individual flows only provides
information about network capacity when the flow is
Network Limited.
Throughput for Data Generation limited flows or for Flow Control limited
flows is related to source and destination resources or
application state.
Evolution of average user throughput on Network Limited flows, showing the decrease in available capacity for individual flows at peak hours on downstream traffic.
Retransmissions and Congestion Events
By following state for all connections Analyzer detects all retransmissions, and also explicit congestion notifications.
Retransmissions percentage evolution for incoming and outgoing traffic.
Smart Routing
The Analyzer can communicate with routing infrastructure
and combine dinamic routing information with performance metrics
to evaluate multiple providers in multi-homing scenarios.
It can also integrate with Palermotec Smart Edge Router or major
router vendors equipment to achieve "dinamic-performance-routing"
in order to boost quality to network content customers.
Report showing performance, routing and geo data.
Round Trip Times
Analyzer measures actual round trip times experienced
by user connections, caused by the interaction of
their congestion control algorithms and their path bottlnecks.
This is extremely useful to detect bufferbloat conditions
and to tune active queues, customer policers, etc.
Two Half Round Trip Times are measured.
Evolution of aggregate average and minimum Round Trip Times for finished connections, showing their variation with network conditions.
Server Profiling
The Analyzer not only measures networks but also server perfomance.
This allows profiling different application response times for
individual use cases. It also allows distinguishing between overloaded
servers and sistematic application delays.
Server average response times list, and indivudual server response time showing characteristic transaction times and load spread.
Taylored and automatic alarms
Analyzer can automatically set alarm thresholds and criteria based on end-user quality of service. They can also be taylored to specific management needs. Alarms encompass end-to-end communications covering broad scopes or fine details of infrastructure and services as needed on each case.
Automatic and taylored alarms on end-to-end communications.
Asymmetric Routing
Network configurations may cause large asymmetric routing volume.
When the Analyzer observes one sided traffic, it can estimate the
unseen traffic volume from the reverse traffic.
This is very useful for route debugging, to anticipate changes in
link traffic volume, and to estimate the costs of bad routing policies
or configurations.
Analyzer observing only one side of asymmetric traffic.
Traffic profile with large percentage of Unseen Traffic, calculated from reverse traffic.