33.网络监控简介

33.网络监控简介

Introduction to Network Monitoring
Dynatrace network monitoring provides insight into the quality of communications between hosts and processes. There is very little overhead generated by Dynatrace network monitoring. If overhead increases above 5% of available CPU, Dynatrace pauses the network module for 3 minutes. If the threshold still exceeded when network monitoring is resumed, throttling occurs again, with the network module paused for twice as long. This continues until the threshold is no longer exceeded up to a maximum pause time of 45 minutes.
To access these options, select the host networking icon. The network overview provides infographics of main key performance indicators or KPIs. Total incoming traffic is compared to outgoing traffic, and connectivity is measured against refused and timed out connections. Retransmissions occur when packets are blocked or dropped due to congestion.
Let’s analyze process connections. Choose a host to expand an infographic that details traffic, retransmissions, and connectivity for that particular group. Select host details to examine incoming and outgoing connections. Returning to the host page, click analyze connections to view them from the individual process and network adapter perspective. By default, process CPU usage and memory utilization are displayed with a list of individual process groups shown by the most frequently used technology.
Drill even further to inspect particular aspects of that specific process group. In this case connectivity and traffic are shown compared to CPU and memory, exposing bottlenecks and slowdowns.
Communications are an important part of traffic flow. Select interfaces to display adapters and quality of network connectivity. Choose an adapter to isolate traffic then network interface details. Scrolling to network analysis details dropdowns provides a full range of information regarding that specific adapter. Dropped packets, errors, refused or dropped connctions can cause hidden problems. These are often shown by high numbers of transmission retries. Statistics are displayed by count, percentages, minimum, maximum and other criteria. These can be shown in the data explorer or pinned to a dashboard.
Network traffic can also be sorted by process. Click the dropdown to select a process to examine. Traffic in and out are shown for the selected process. Retransmissions show repeated attempts to make the connection. Similarly, conn ections that are refused or time out can cause errors. Handshake Round Trip Times measured during TCP session establishment are the most precise RTT measure. This measure is especially useful when clients from remote locations or hosts in different availability zones are involved. Disk throughput examines disk writes and reads per second and can show slowdowns and interruptions. Click on process details to show the impact of a particular service on network connectivity.
For more about these options, please refer to the Dynatrace documentation.