Diagnosis and Performance Monitoring - Deploying and Managing a Cloud Infrastructure - page 136

interactive applications like online gaming as well as for examining app-to-app situations
such as measuring how fast two servers like a web server and a database server can inter-
act with each other and exchange data.

FIGURE 5.4 Simple representation of round-trip time

Transmission time

S

Network lag

Transmission time

Round-
trip
time

Processing lag

E

Source

Network
intermediary

Destination

Jitter Jitter is the variation in packet traversal or transmission delay that is caused by
queuing and is the effect of contention or conflict and serialization (where data packets can
travel only one after the other) that takes place anywhere on the network path between and
within networks.
End Point Computational Speed This simply refers to how fast the server, client, or what-
ever end points that are communicating can process data that has been received and how
quickly they can send an appropriate reply. This is dictated by core computational elements
of the device: the CPU, memory, and other core hardware and software.
Complexity can add to latency. The number-one source of latency is distributed computing.
The nature of applications has changed, and now a lot of applications, especially those related
to data analytics and big data, employ hundreds if not thousands of separate servers world-
wide, which all have varying degrees of latency brought by the Internet connectivity and the
networks between them. Depending on the time that these servers are running, the condition
of network congestion in their respective areas could be good or bad as Internet traffic waxes
and wanes as competition for bandwidth and infrastructure take place.
Virtualization is another source of complexity that adds to overall latency. Virtual
networks and virtual desktops all mimic various hardware, including network interfaces,
but they still have to be translated into hardware signals from the software virtualization.
This adds another layer of processing latency on top of the network latency.

Hop Counts

Hop count is literally the number of hops that a data packet will take from one network
node to the next until it reaches its destination. For example, if there are three routers