Chapter 8 - Performance

Question 1

what is performance?

Answer 1

the ability of a system to meet timing requirements

Question 2

for the performance general scenario, what is the source

Answer 2

internal or external to the system

Question 3

for the performance general scenario, what is the stimulus

Answer 3

arrival of a periodic, sporadic, or stochastic event

Question 4

for the performance general scenario, what is the artifact

Answer 4

the system or one or more of its components

Question 5

for the performance general scenario, what are the 4 possible environments

Answer 5

normal mode

emergency mode

peak load

overload mode

Question 6

for the performance general scenario, what are the 2 responses

Answer 6

process events

change level of service

Question 7

for the performance general scenario, what are the 5 possible response measures

Answer 7

latency
deadline
throughput
jitter
miss rate

Question 8

what is the primary goal of performance tactics

Answer 8

to generate a response to an event arriving at the system within some time-based constraint

Question 9

2 categories of performance tactics

Answer 9

control resource demand

manage resources

Question 10

6 tactics for controlling resource demand

Answer 10

• manage sampling rate
• limit event response
• prioritize events
• reduce overhead
• bound execution times
• increase resource efficiency

Question 11

what does it mean to manage sampling rate

Answer 11

to reduce the sampling frequency at which a stream of data is captured if possible typically with some loss of fidelity

Question 12

what does it mean to limit event response

Answer 12

process events only up to a set maximum rate to ensure more predictable processing

Question 13

what does it mean to prioritize events

Answer 13

you can rank events in order of how important it is to service them

Question 14

what does it mean to bound execution times

Answer 14

place a limit on how much execution time is used to respond to an event

Question 15

what does it mean to increase resource efficiency

Answer 15

improving algorithms used in critical areas to decrease latency

Question 16

6 tactics for managing resources

Answer 16

increase resources

introduce concurrency

maintain multiple copies of computations

maintain multiple copies of the data

bound queue sizes

schedule resources

Question 17

what does it mean to maintain multiple copies of computations

Answer 17

have replicas on different servers to reduce resource contention that would occur if all computations took place on a single server

Question 18

what does it mean to maintain multiple copies of data

Answer 18

to keep copies of data on storage media with different access speeds

Question 19

what does it mean to bound queue sizes

Answer 19

control the maximum number of queued arrivals and consequently, the resources used to process the arrivals

Question 20

what does it mean to schedule resources

Answer 20

when there is resource contention, the resource must be scheduled

Question 21

2 issues with performance in regards to allocation of responsibilities

Answer 21

• determine the system responsibilities that will involve heavy loading, have critical time-response requirements, or impact portions of the system where heavy load or time-critical stuff occurs
• ensure required performance response can be met

Question 22

5 things to consider for system responsibilities that involve heavy loading or have time-critical response requirements

Answer 22

• identify process requirements of each responsibility to determine if any bottlenecks will occur
• identify responsibilities to recognize and process requests appropriately
• identify threads that cross process or processor boundaries
• identify responsibilities to manage threads
• identify responsibilities for scheduling shared resources

Question 23

4 things to consider when evaluating a coordination model with respect to performance

Answer 23

• does it support concurrency, event prioritization, or a scheduling strategy
• does it ensure that required performance response can be delivered
• can it capture periodic, stochastic, or sporadic event arrivals
• does it have the appropriate properties of the communication mechanism

Question 24

4 things to consider for data abstractions of the data model that will be heavily loaded or have time-critical response requirements

Answer 24

• determine whether keeping copies of key data would benefit performance
• determine if partitioning data would benefit performance
• determine whether reducing processing requirements for data abstractions is possible
• determine whether adding resources to reduce bottlenecks is feasible

Question 25

4 issues with performance in regards to mapping among architectural elements

Answer 25

- determine where heavy network loading will occur and whether or not to co-locate some components - ensure that components with heavy computational requirements are assigned to processors with the most processing capacity - determine where introducing concurrency can have a significant positive effect - determine whether the choice of threads and the responsibilities introduce bottlenecks

Question 26

2 issues with performance in regards to resource management

Answer 26

- determine whether resources in the system are critical for performance - for such resources ensure that they will be managed under normal and overloaded system operation

Question 27

3 issues with performance in regards to binding time

Answer 27

- determine the time necessary to complete any bindings after compile time - determine the additional overhead introduced by using the late binding mechanism - ensure that the above values do not impose unacceptable performance penalties on the system

Question 28

4 issues with performance in regards to choice of technology

Answer 28

- will the choice of technologies let you set and meet hard real-time deadlines - do you know its characteristics under load and its limits - does your choice of technology introduce excessive overhead for heavily used ops