WN - 3.05 Flashcards
Describe the poisson process
Stochastic process in which events occur continuosly and independently of one another
Explain burstiness
A bursty source generates traffic in random clusters
Is deterministic traffic bursty or not bursty?
bursty
What is the poisson process in discrete time
Bernoulli process
When is the burstiness of poisson and bernoulli processes removed
Removed for aggregated traffic sources
What is self-similar traffic?
Maintains its burstiness at any time scale
Why is burstiness important?
- Peak traffic demands on bugger resources can lead to overflow and lost traffic
- Peak demands may create QoS problems in a network
- Need to characterise burstiness for traffic sources in a QoS environment
What are the properties of the self-similar phenomena?
- Have structure at arbitrarily small scales
- A self-similar structure contains smaller replicas of itself at all scales
- For real phenomena, properties do not hold indefinitely; however, they hold over a large range of scales
What is the Hurst parameter
A key measure of self-similarity
When H=0.5 -> indicates the absence of self-similarity
H closer to 1 indicates a higher degree of persistence of long-range dependance
What is data traffic well-modelled as?
A self-similar process in many practical networking situations
- Ethernet traffic
- WWW traffic
- TCP, FTP traffic
VBR video
Straightforward queuing analysis using poisson traffic assumptions inadequate to model this type of traffic
In regards to ethernet traffic, what happens when the load increase
The Hurst parameter increases as well
What does aggregating several streams do?
It does not remove self-similarity
Poisson is not a good model in this case. What is?
Superposition of many Pareto-like ON/OFF sources
Elaborate on the performance implications
- If actual data is more bursty than originally modeled, then the original models underestimate average delay and blocking
- Self-similarity leads to higher delays and higher blocking probabilities
- Therefore, self-similarity leads to a poor fit with traditional queuing theory results