Database Reference
In-Depth Information
Table 4.1
Building a test set
Test set method
Test scenario
Load test
High stress round-trip
End-user-cloud host
High stress test strategy
Low stress database read and write
Cloud host-cloud database
Low stress test strategy
High stress database read and write
Cloud host-cloud database
High stress test strategy
Low stress large file read, write, and delete
End-user-cloud database
Low stress test strategy
High stress round-trip
: The end-users concurrently send message requests to
cloud hosting servers. For each request received, the servers immediately echo
back to the end-users with the received messages. The response time is recorded
in this test. This is the base test that provides a good benchmark for a total round
trip cloud application usage experience as the response time as experienced by
the average end-user will be affected by the various variable network conditions.
This is a useful test to indicate the likely end-user experience in an end-to-end
system testing scenario.
Low stress database read and write
uses the cloud host-cloud database scenario.
It starts with the low stress test strategy, which provides an initial reference result
set for subsequent high stress load tests. This test is performed with varying data
sizes, representing different cloud application data types. The data types provided
by the CARE framework are: a single character of 1 byte, a message of 100
bytes, an article of 1 KB, and a small file of 1 MB. These data types are sent
along with the read or write requests, one after another to the cloud databases
via the cloud hosting servers. The database processing time will be recorded and
then returned to the end-user within the response. In terms of request size the
CARE framework follows the conventional cloud application design principle
of storing data that are no larger than 1 kB in structured data oriented storage,
namely Amazon SimpleDB and Microsoft Windows Azure Table Storage. Data
that are larger than 1 KB will be put into binary data oriented databases, including
Amazon S3 and Microsoft Windows Azure Blob Storage. In addition, Google
App Engine Datastore supports both structured data and binary data in the same
cloud database.
High stress database read and write
are based on the high stress test strategy.
It simulates multiple read/write actions concurrently. The number of concurrent
requests range is configurable, as described in Sect.
4.1
. Due to some common
cloud platform quota limitations, for example Google App Engine by default
limits incoming bandwidth to a maximum of 56 MB/min, this test uses a default
test data size of 1 kB. This test data size can be configured to use alternative
test data sizes if the target testing cloud platform does not have those quota
limitations. Lastly, a cron job is scheduled to perform the stress database test
repeatedly over different time periods across the 24 h period.
