Which metric would you compare to assess whether a load balancer distributes traffic evenly across backend servers?

• A. Total number of conversations
• B. Bytes per backend server
• C. DNS TTL values
• D. Average packet size

Answer: (B)

Distributing load evenly means you want to see that each backend is handling a similar amount of work. The most direct way to measure that is by looking at the data volume each backend processes—bytes per backend server. If one server handles a lot more bytes than the others, the traffic isn’t being balanced evenly, which could indicate a configuration issue, session stickiness effects, or uneven traffic patterns.

Why the other metrics aren’t as reliable for this purpose: counting how many conversations (connections) there are doesn’t tell you how much data those connections move. You could have the same number of connections but one server carrying much heavier payloads. DNS TTL values affect how long a resolver caches a record and don’t measure how traffic is distributed among backends. The average packet size can be informative for network characteristics, but it doesn’t reveal how traffic is split across servers; different servers could see the same average while one handles far more total data due to more connections or heavier payloads.

So, bytes per backend server gives a direct view of load distribution across the pool, making it the best metric to assess evenness of traffic distribution.