Page 58 - 130902

EducationContinueducucatatioioionnContinued
Ed
Ed
Ed
Stub, local loopback, processorb, local loopback, processor
Stu should remain within a narrow band; that is, low standard hould remain within a narrow band; that is, low standard
s
deviation. For example, it may be acceptable that 95 y be acceptable that 95
deviation. For example, it ma
P p
Performance test cases can employ a local loopback erformance test cases can employ a local loopback
percent of the transactions complete their round-trip ercent of the transactions complete their round-trip
i
inside the gateway, a stub representing a processor or nside the gateway, a stub representing a processor or j journey inside the gateway within one second. Though 5 ourney inside the gateway within one second. Though 5
an actual processor. The most realistic scenario is to processor. The most realistic scenario is to
an actual percent do not fall in this boundary, it does show a high
percent do not fall in this boundary, it does show a high
r
run the test cases with a processor, but that isn't always un the test cases with a processor, but that isn't always degree of gateway performance.
degree of gateway performance.
possible. Many processors throttle the performance in ossible. Many processors throttle the performance in
p
t their test systems to enable all gateways to test without heir test systems to enable all gateways to test without Statistical mode represents the most common value in the
Statistical mode represents the most common value in the
performance degradation.erformance degradation. collection of data points. This is the most likely value that
p
collection of data points. This is the most likely value that
occurs, and hence most data in the collection have this
occurs, and hence most data in the collection have this
An a
An approach to resolve this issue is to use a processor pproach to resolve this issue is to use a processor value. For example, if the mode of transaction round trip is
value. For example, if the mode of transaction round trip is
stub t
stub that will use the same network connectivity, hat will use the same network connectivity, 0.8 seconds in a collection of values for transaction round
0.8 seconds in a collection of values for transaction round
accept incomin trip, it indicates most merchant transactions are handled in
accept incoming transactions and send back a g transactions and send back a
trip, it indicates most merchant transactions are handled in
r
response. Advanced stubs can even vary the response esponse. Advanced stubs can even vary the response 0.8 seconds. Some transactions may be handled with even
0.8 seconds. Some transactions may be handled with even
at random or on a pr better performance; others may be worse.
at random or on a predefined basis – denying every edefined basis – denying every
better performance; others may be worse.
f fifth transaction, for example.ifth transaction, for example.
Testing strategy recommendationsesting strategy recommendations
If T
If writing a stub isn't an option, some developers writing a stub isn't an option, some developers
use a local loopback method via which a transaction se a local loopback method via which a transaction
u
A
r
response is generated inside the gateway itself. While esponse is generated inside the gateway itself. While A test platform built for payment gateway performance test platform built for payment gateway performance
testin
this method is simplest to implement, it does not testing should capture quite a few metrics, some of which I g should capture quite a few metrics, some of which I
this method is simplest to implement, it does not
have just been mentioned. While test strategy is governed ave just been mentioned. While test strategy is governed
reflect the reality of transaction processing, except for h
reflect the reality of transaction processing, except for
by the overall expectations and goals, here are four best the overall expectations and goals, here are four best
stand-in processing scenarios.
stand-in processing scenarios. by
practices to follow:ractices to follow:
p
Failover scenarioslover scenarios
Fai 1. Generally, do a performance test repeatedly for lly, do a performance test repeatedly for
1. Genera
c comparison purposes. To accommodate this, I suggest omparison purposes. To accommodate this, I suggest
Failover handling puts an additional burden on ailover handling puts an additional burden on c creating an automated test harness at the beginning reating an automated test harness at the beginning
F
t the payment gateway. During such times, some he payment gateway. During such times, some and maintaining it wit
and maintaining it with every change to the code base, h every change to the code base,
transactions ar
transactions are either slowed down or dropped, e either slowed down or dropped, updating as needed. Test cases must be predefined pdating as needed. Test cases must be predefined
u
r resulting in timeouts. While these are important esulting in timeouts. While these are important and upgraded from time to time to reflect changing graded from time to time to reflect changing
and up
scenarios, they are not daily events. It's essential to run hey are not daily events. It's essential to run
scenarios, t business scenarios. usiness scenarios.
b
p
performance tests that include and exclude failover erformance tests that include and exclude failover 2. A clear understanding of the transaction mix is A clear understanding of the transaction mix is
2.
scenarios and understand the results in each case.os and understand the results in each case. essential. Design the performance test so that it reflects ssential. Design the performance test so that it reflects
scenari
e
t the target transaction mix.he target transaction mix.
3. Capture and store multiple data points for future pture and store multiple data points for future
Stress test
Stress testinging 3. Ca
c comparisons. Raw data points must be retained so that omparisons. Raw data points must be retained so that
f
T future calculations can be handled.uture calculations can be handled.
The most common reason for performance testing is to he most common reason for performance testing is to
4. Apply appropriate statistical calculation to the data to . Apply appropriate statistical calculation to the data to
know the platform's capacity during stress situations. he platform's capacity during stress situations.
know t 4
draw accurate conclusions fr
I It is important to know how the gateway is performing t is important to know how the gateway is performing draw accurate conclusions from the testing.om the testing.
under high load conditions, as well as how the nder high load conditions, as well as how the
u
Gateways are complex systems, and businesses have ys are complex systems, and businesses have
hardware handles disk usage, queuing, memory usage ardware handles disk usage, queuing, memory usage
h Gatewa
high expectations for them. Evaluating performance
and swap, network utilization, etc. A stress test allows
and swap, network utilization, etc. A stress test allows high expectations for them. Evaluating performance
requires precision. There are a number of moving parts
for future capacity planning both in terms of software
for future capacity planning both in terms of software requires precision. There are a number of moving parts
in the evaluation process, test methods and analysis of
tuneup and hardware upgrades.
tuneup and hardware upgrades. in the evaluation process, test methods and analysis of
the methods. A repeatable, automated performance test
the methods. A repeatable, automated performance test
suite should be a mandate for any quality assurance team
suite should be a mandate for any quality assurance team
Statistical conceptstatistical concepts
S responsible for payment gateway testing. A planned
responsible for payment gateway testing. A planned
approach to performance testing with clarity on strategy,
approach to performance testing with clarity on strategy,
While performance testing can generate valuable data, hile performance testing can generate valuable data,
W
goals and methods can lead to enhanced payment gateway
i interpreting that data requires some knowledge of nterpreting that data requires some knowledge of goals and methods can lead to enhanced payment gateway
quality and stability.
statistical concepts. Arithmetic mean is useful to find out pts. Arithmetic mean is useful to find out
statistical conce quality and stability.
avera
average values such as transactions per second, but a mean ge values such as transactions per second, but a mean
C
does not reflect the spread of the data points.eflect the spread of the data points. Chandan Mukherjee is the co-founder of PayCube Inc., which is a handan Mukherjee is the co-founder of PayCube Inc., which is a
does not r
San Francisco Bay Area, Calif.-based payment consulting and IT
San Francisco Bay Area, Calif.-based payment consulting and IT
T
The spread of data points is better reflected in variance or he spread of data points is better reflected in variance or services company providing custom software solutions and custom
services company providing custom software solutions and custom
standar
standard deviation. This reveals the spread of the data. An d deviation. This reveals the spread of the data. An gateways for acquirers, ISOs, retailers and varied organizations
gateways for acquirers, ISOs, retailers and varied organizations
e in the world of payments and consumer transaction acquiring and
example is the system's total throughput. If the data points xample is the system's total throughput. If the data points
in the world of payments and consumer transaction acquiring and
h management, including prepaid and gift card program, loyalty
have low variance, it will imply that most transactions ave low variance, it will imply that most transactions
management, including prepaid and gift card program, loyalty
h and promotion, payment start-up, POS solution, mobile payment
have similar performance. ave similar performance.
and promotion, payment start-up, POS solution, mobile payment
and e-commerce players. For more information, email cm@pay-
and e-commerce players. For more information, email cm@pay-
F cubeinc.com, call 510-545-6854 or visit www.paycubeinc.com.
For most performance-related measurements, the value or most performance-related measurements, the value
cubeinc.com, call 510-545-6854 or visit www.paycubeinc.com.
588
5