Is it the network?
The Network Performance dashboard helps you determine if network roundtrip and latency are impacting server response time to users. It focuses on network-centric information like network latency and delay, network roundtrip, device byte counts, retransmissions, and utilization.
Select the Network Performance dashboard. To get the dashboard, see Importing a user-contributed dashboard.
Let’s look at some widgets on this dashboard from a “slow server” perspective.
Total Server Response Time is calculated by the difference between the server acknowledging data receipt and the first byte to the server reply. Network delay is a combination of network delays calculated for both the client and server side. It is calculated first by looking at the TCP three-way handshake. Throughout the rest of the conversation additional measurements for network delay are taken by looking at the difference in time between every two received data packets and the acknowledgment of that data. If a faster acknowledgment is seen on either side, the network delay is updated to include the lower calculated delay.
Let’s first look at a healthy network by using our dashboard widget for SQL servers. The network delay roundtrip metric is always a good starting point. The Server Network Delay- Round Trip widget shows server network delay metrics sorted by the longest round trip time. If that number is low, then your packets are moving efficiently through the network and the latency is low. Notice our round trip, maximum, and average delays are all low. This network is running well. We don’t see network delay, latency-centric problems, or many retransmissions here.
 
Figure 51: Healthy network
 
Now let’s take a look at an unhealthy network using the same widget, but focused on HTTPS servers. Our first HTTPS server, 52.109.124.20, has an average Network Delay Round Trip time of 234.94 ms. The table shows network delay by direction: from client to server, from server to client, as well average and maximum values. Knowing the direction of the delay is helpful.
It’s also important to compare the average and maximum value. If the average value is high it suggests this problem is happening all the time. If the average is low but the maximum is high, it suggests that this may have been a temporary condition or just an anomaly, perhaps caused by a spike in network utilization. Also note that retransmissions are 0, so we know packets are not being lost. The servers in this table are showing high average and maximum network delays which suggests that the network is not running efficiently, this is indicative of a network issue.
 
Figure 52: Unhealthy network
 
To further troubleshoot round trip/latency issues, it’s helpful to know additional information, such as what does the network topology look like, where is the server located, are users accessing the server over a WAN, what is the utilization of the various network segments being traversed. It’s also important to know where the probe is located in the network for context of the network delay metrics. The network delay round trip value is usually the best metric to quickly assess network latency problems.