The Refractor uses data files collected by the Resolver Tool
but automatically performs computations relating each of the tiers
in a distributed application. The data collected by the Resolver Tool at each
tier in an application can be combined with the other tier data to
gain an understanding of the overall response time of all tiers combined
into a single plot. The workload analysis performed by Refractor has an
initial focus on client-server
and web based application architectures but also includes a planning
application from i2. In the future, additional workload types will
be added.
To select a file of data for analysis the analyst uses the file browser
and selects a file to be read as shown in Figure 1.
Figure 1
In order to relate the applications and hosts by tier, the analyst must
select hosts and processes in each tier before the computations can be
performed unless there is an automatic workload builder. The selection
panel is shown in Figure 2 below. For this particular data file, we know
that the workload was WebSphere so the analyst has chosen Webshpere as
the workload to be built. In the case of the manual selections, the
analyst would select among the pids and hosts in the data file to build
up the processes in each tier. Once that is complete the data can be
analyzed as we showed in the previous example.
Figure 2
