eG Monitoring
 

Measures reported by RADLBRSStatTest

Real servers are dedicated physical servers that you typically configure in groups called real server groups. The Radware Alteon Load Balancer distributes incoming network traffic and session load across a real server group.

Typically, the real servers in a group are bound to a virtual server. Whenever a client request is received, the virtual server bound to a real server group responds to those requests by routing the requests to those real servers in that group that are currently available. Load distribution can be done in different ways. The load balancer may use a round-robin method, where each server is used in turn. It can also use a weighted round robin system, where servers are assigned traffic based on their configured capabilities. Regardless of which load balancing technique/algorithm is used, the aim is to ensure that no single real server is overloaded with requests.

If one/more real servers in the group are unavailable, then the load should be shared by the other real servers in the group. However, if at any given point in time, one/more real servers in a group handle a significantly higher session and/or data load than the rest, it is a clear indicator of ineffective load-balancing! Under such circumstances, administrators may have to fine-tune/change the load-balancing algorithm.

To be able to quickly and accurately spot load-balancing irregularities and to initiate remedial measures, administrators should keep a close watch on the availability of each real server in a group and the session and data load handled by each server. This is what the Real Servers Test does!

For each real server in a group, this test reports the availability and operational status of the server. In addition, the test also enables you to analyze the impact of the unavailability (if any) of a server on the load imposed on other servers in the group by reporting the total session load on each server and data handled by every server in the group. This way, the test sheds light on issues in load-balancing, and thus urges administrators to take appropriate corrective action.

Outputs of the test : One set of results for each real server in every server group that is mapped to a virtual server.

Descriptor of the test : Real Server

The measures made by this test are as follows:

Measurement Description Measurement Unit Interpretation
rsStatus Indicates the current status of this real server.   The values that this measure can report and their corresponding numeric values are as follows:

Measure Value Numeric Value
Running 2
Disabled 4
Failed 5

Note:

By default, this measure reports the Measure Values discussed in the table above to indicate the status of the real server. In the graph of the measure however, the real server status is indicated using the corresponding numeric equivalents only. Use the detailed diagnosis of this measure to determine the IP address of teh real server.
curSession Indicates the number of sessions currently handled by this real server. Number Compare the value of this measure across real servers in a particular group to determine whether/not session load has been uniformly distributed across the servers in that group. If load distribution is not uniform, then identify which real server has handled a significantly higher session load than the rest. You will then have to investigate the reasons for this load-balancing faux pas - is it because the other servers in the group are unavailable? If this is not the case, then you may have to fine-tune the load-balancing algorithm.
totalSession Indicates the total number of sessions handled by this real server from the time it was started. Number  
rsFailure Indicates whether/not this real server was down even once during the last measurement period.   The values that this measure can report and their corresponding numeric values are as follows:

Measure Value Numeric Value
Yes 1
No 0

Note:

By default, this measure reports the Measure Values discussed in the table above to indicate the status of the real server. In the graph of the measure however, the real server status is indicated using the corresponding numeric equivalents only.
transmitRate Indicates the rate at which data was handled by this real server. MB/Sec Compare the value of this measure across the real servers in a group to figure out if any particular server is handling a significantly higher data load than the others in the group. If so, this could hint at a load-balancing irregularity. You may want to investigate the reasons for this-could it be because the other servers are unavailable? If this is not the case, you may want to fine-tune the load-balancing algorithm.