WAN optimization adjustment in NAM is designed to make your deployment aware of WAN optimization in the network and able to rectify measurements that might be skewed by optimization.
NAM is capable of collecting and reporting numerous metrics based on monitored network traffic within your network infrastructure. When these metrics are used in an optimized WAN environment, however, their calculations must be adjusted to account for the presence of WAN optimization. NAM does so by detecting optimized traffic on WAN links and recalculating network metrics respecting WAN optimization.
NAM Probe deployment
The typical deployment, in which a NAM Probe is connected on both sides of the data center WOC, makes it possible to observe the network traffic before and after it is optimized. This deployment requires the ports destined for WAN optimization to be spanned on both sides of the WOC so the traffic destined for optimization and traffic already optimized is available to the NAM Probe. With this information, NAM is able to correlate WAN packets and adjust metrics by utilizing the Application Delivery Channel Delay (ADCD) metric.
Figure 1. Single NAM Probe Monitoring Both Sides of a WAN Optimization Controller
The transaction measurements performed on an unoptimized link are adjusted based on measurements taken on the optimized link. In protocols involving more than one operation in a single connection, the NAM Probe correlates optimized packets with operations.
Using this single NAM Probe configuration, NAM performs the following adjustments:
Correlate WAN packets The report server correlates WAN packets belonging to a specific transaction.
Introduce the Application Delivery Channel Delay metric. If protocol or application traffic does not allow for packet correlation, Dynatrace uses the ADCD metric to adjust network and WAN metrics.Note
WAN Optimization metrics added to the standard DMI report may alter the values of certain metrics. For example, a report displaying
Total bytesfor a specific software service, will display a different value for the same
Total bytesmetric if that report contains metrics associated with optimized WAN. Adding a WAN Optimization metric to the report widens the perspective of that report and as such, the report then displays the WAN Optimization metrics combined with values observed locally. As a result, a duplication of some metric values occurs.
To configure your deployment to monitor WAN optimized traffic, go to:
NAM Console ► NAM Probe Configuration ► Global ► WAN Optimization
The in-path deployment requires that the WOC be connected to the LAN-side device (datacenter backbone switch) and the corresponding WAN-side device (router).
WAN acceleration engine vendor allows you to select your deployment vendor:
WAN optimization support has been thoroughly tested against physical Steelhead appliances. In addition, tests against Steelhead Mobile, a software equivalent of the WOC for remote users, have also been conducted, but in a narrower range.
- Physical in-path.
- Virtual in-path WCCP.
Network Application Monitoring supports usage of Web Cache Communications Protocol (WCCP) where Riverbed Steelhead appliances are connected to a switch in the path of the network flow between the servers and the clients.
Network Application Monitoring also supports deployments in which the Riverbed Interceptor redirects traffic to a cluster of Riverbed Steelheads. This deployment requires that the Riverbed Interceptor device be deployed with access to data as if it was in-path (for example, the traffic flow can be mirrored to the NAM Probe simulating the in-path deployment); otherwise, NAM would not be able to provide correct measurements.
WAN optimization support has been thoroughly tested against physical Cisco WAAS appliances. The following deployments are supported:
- Virtual in-path WCCP (default deployment for WAAS).
- Physical in-path, also known as inline.
A NAM Probe monitoring all network traffic at various monitoring points within your infrastructure needs to distinguish the types of network traffic that it is observing. You can classify the traffic types by identifying the networks in the monitoring point rules. The three network types that you can identify within your WAN optimization deployment are:
- LAN Optimized
- Datacenter LAN
Depending on your deployment model, you may be required to identify all three networks. You can identify the networks based on network interface, VLAN, MPLS, or GRE/WCCP values. The more network traffic you identify and classify by type, the more precise the analysis of your WAN optimization performance will be.
- In-path single point This monitoring deployment assumes that the NAM Probe monitors the network traffic before the WOC, and therefore that identification of optimized WAN traffic is required. After the specific network traffic on the WAN side has been identified as WAN optimized, the rest of the network traffic (the LAN side) is automatically classified as regular LAN traffic (Datacenter LAN). However, traffic on the LAN side that is matched with the traffic on the WAN side is automatically reclassified as LAN traffic destined for, or returning from, optimization (LAN optimized). This configuration requires minimal knowledge of the traffic that is being observed, but the LAN side relies on automatic detection and matching of optimized WAN and LAN traffic, so this configuration may be inadequate for certain applications.
- In-path double point (LAN optimized) Defining the network traffic type on the LAN side, you can specify the traffic that you expect to be, or already was, optimized (LAN optimized). This, along with matched traffic on the WAN side, will present the performance data of your WAN optimization environment. All other traffic observed on the LAN side is classified and reported as regular LAN traffic (Datacenter LAN).
- In-path double point (LAN) Use a configuration like this to separate optimized LAN traffic from the rest of your LAN traffic. This can be achieved by specifying regular LAN traffic. This configuration isolates a segment of your LAN that you do not expect to be optimized; its traffic, therefore, does not need to be correlated with the optimized traffic on the WAN side. This makes it possible to identify the traffic relating only to clients utilizing the connections that are not optimized and at the same time to identify the rest of the traffic as relating only to clients using the connections that are optimized.
- In-path advanced mode
Classifying all possible network traffic enables more precise measurements, but it requires more information about the monitored environment. The three types of networks that should be identified in the monitoring point rules are:
- LAN Optimized
- Datacenter LAN
The advanced mode also allows you to specify the network traffic that should not be taken into consideration at all and should be omitted in WAN optimization performance calculations. The network identification defined in the monitoring point rule as type “
Drop” will not be monitored as part of WAN optimization.
- Virtual in-path advanced mode
When virtual in-path deployment is in use, WAN traffic is often redirected by the routing device and the WOC device. It is important to identify traffic belonging to the WAN optimized network and separate it from all other traffic. Typically, the WAN optimized network is identified by VLAN, MPLS, or GRE/WCCP values in the monitoring point rules, and the following rule would set all other traffic on that interface as “
Monitoring point rules
You can define up to 16 rules for your deployment choice. To define a rule, right-click on the rules table and select Add. The rules are processed from top to bottom. Any of the real in-path deployment configurations can be copied to In-path advanced mode. Clicking the Save or the Save and Publish button will accept and propagate the currently selected deployment and all the monitoring point rules.
In the advanced mode deployment, where no automatic traffic classification takes place, after all of the monitoring point rules have been applied, the remaining network traffic can be manually classified.
All other traffic
The Treat all other traffic as setting allows you to specify how the rest of the observed network traffic should be classified.