Database Reference
In-Depth Information
adaptive SLA management framework for the different experiments fall between
the two baselines based on the variance on the monitor interval, int
v
l
mon
, and the
tolerance of replication delay, delay
tolerance
. However, it is worth noting that the
end-to-end throughput can be still affected by a lot of performance variations in
the cloud environment such as hardware performance variation, network variation
and warm up time of the virtualized database servers. Similarly, The two baseline
experiments also represent the minimum and the maximum running time of all
virutalized database replica servers with 9,000 and 18,000 s respectively. Therefore,
the total running time of the replica servers for the different experiments fall within
the range of 9,000 and 18,000 s. Each experiment starts with 3 active replicas which
are gradually increased during the experiments based on the configurations of the
monitor interval and the SLA of replication delay parameters until it finally ends
with 6 replica servers.
In general, the relationship between the running time of all slaves and end-to-end
throughput is not straightforward. Intuitively, a longer monitor interval or a longer
tolerance of replication delay usually postpones the addition of new virtualized
database replica servers and consequently reduces the end-to-end throughput. The
results show that the tolerance of the replication delay parameter, delay
tolerance
is
more sensitive than the monitor interval parameter, int
v
l
mon
. For example, setting
the values of the tolerance of the replication delay to 4,000 and 1,000 result in longer
running times of the replica servers than when the values are set to 2,000 and 500.
On the other hand, the increase of running time of all replica servers clearly follows
a linear trend along with the increase of the end-to-end throughput. However, a
general conclusion can not be made as the trend is likely affected by the workload
characteristics.
Replication Delay
Figures
7.4
-
7.12
illustrate the effect of the adaptive SLA management framework
on the performance of the replication delay for the virtualized database replica
servers. Figures
7.4
and
7.5
show the replication delay of the two baseline cases that
will be used for comparison purposes. They represent the experiments of running
with a fixed number of virtualized database replica servers, 3 and 6 respectively,
from the start until the end of the experiments. Figure
7.4
shows that the replication
delay tends to follow different patterns for different replica servers. The two trends
of virtualized database servers in
us-west-1
and
eu-west-1
surge significantly at 260
and 280 users respectively. At the same time, the trend of virtualized database server
in
us-east-1
tends to be stable through out the entire running time of the experiment.
The main reason behind that is the performance variation between the virtualized
database servers for replicas, as both virtualized database servers in
us-west-1
and
eu-west-1
are powered by Intel(R) Xeon(R) E5507 @ 2.27 GHz CPU, whereas the
server in
us-east-1
is deployed with a higher performance CPU, Intel(R) Xeon(R)
E5645 @ 2.40 GHz CPU. Due to the performance differences between the physical
