Specifications

2
Principled Technologies, Inc.: Virtualized Exchange workload performance comparison of end-to-end
solutions with iSCSI storage connections: Dell PowerEdge 2950 with Dell EqualLogic PS5000 vs. HP
ProLiant DL385 with HP Stora
g
eWorks EV
A
4400
Workload
Our goal was to see how many virtual machines running this workload on Microsoft Server 2008 with Exchange
2007 the server could handle with acceptable performance. To reach that goal, we created a new custom
workload that reflected existing standards.
To build that workload, we used Microsoft Exchange LoadGen 2007, an industry-standard tool for benchmarking
an Exchange Mail Server. We used as the basis of that workload the LoadGen settings from an industry-standard
virtualized mail benchmark that uses the Microsoft Exchange Server 2003 MAPI Messaging Benchmark 3
(MMB3). MMB3 is the previous generation Exchange benchmark, which Microsoft has replaced with LoadGen
2007. We modified the MMB3 workload so it would work on the latest version of LoadGen 2007 by using the
Custom feature of that tool. During the tests, LoadGen performs tasks to simulate a standard user generating mail
activity. When the workload finishes, LoadGen reports the response time, which is the number of seconds
necessary to complete the task.
We ran this workload on all virtual machines simultaneously for 30 minutes. We defined a solution to have
delivered acceptable performance if it performed each of the test Exchange operations in less than a second in all
of the VMs.
Test results
Figure 3 shows the total number of virtual machines each solution could handle during testing. For an acceptable
score, a solution had to complete all tasks in the LoadGen workload in 1000 ms (1 second), in all virtual machine
LoadGen tests.
To determine the maximum number of simultaneous virtual machines each solution could handle, we added
virtual machines until one or more of the VMs had a response time on some operations of greater than one
second. We then subtracted one VM to identify the total number of VMs the solution could handle. We performed
three runs at the total number to verify the response time always remained below 1000ms.
Figure 2: Performance/watt results for the servers normalized to that of the HP ProLiant DL385
server and HP StorageWorks EVA 4400 storage array solution. Higher numbers are better.
0.00
0.20
0.40
0.60
0.80
1.00
1.20
1.40
Normalized comparison
-
Performance/watt results
HP ProLiant DL385
server and HP
Sto rageWorks EVA 4400
storage array
Dell PowerEdge 2950
server and Dell
EqualLogic PS5000
storage array