Database Reference
In-Depth Information
Cloudstone benchmark
L1
M / N
satisfies pre-defined read/write ratio
M
write
operations
L2
Master
Replication within the same region and the same availability zone
Slave
1
Slave
k
Slave
k+1
Slave
n
Replication within the same region but across availability zones
L3
Slave
1
Slave
k
Slave
k+1
Slave
n
Replication across regions
Slave
1
Slave
n
Slave
1
Slave
n
Slave
1
Slave
n
Slave
1
Slave
n
us-west
eu-west
ap-southeast
ap-northeast
Fig. 6.1
The architecture of relational database as a service benchmark application
and creating events, as well as, social operations such as joining and tagging
events [
210
]. Unlike Web 1.0 applications, Web 2.0 applications impose many
different behavioral demands on the database. One of the differences is on the write
pattern. As contents of Web 2.0 applications depend on user contributions via blogs,
photos, videos and tags. More write transactions are expected to be processed.
Another difference is on the tolerance with data consistency. In general, Web 2.0
applications are more acceptable to data staleness. For example, it might not be a
mission-critical goal for a social network application like Facebook to immediately
have a user's new status available to his friends. However, a consistency window
of some seconds or even some minutes would be still acceptable. Therefore, it is
believed that the design and workload characteristics of the Cloudstone benchmark
is more suitable to the purpose of the study rather than other benchmarks such
as TPC-W [
49
]orRUBiS[
42
] which are more representative of Web 1.0-like
applications.
The original software stack of Cloudstone consists of three components: web
application, database, and load generator. Throughout the benchmark, the load
generator generates load against the web application which in turn makes use of
the database. The benchmark has been designed for benchmarking the performance
of each tier for Web 2.0 applications. However, the original design of the benchmark
limits the purpose of the experiments by mainly focusing on the database tier
