5
Principled Technologies, Inc.: Dell PowerVault MD3000i high-availability testing on
Windows Server 2008 R2 and Hyper-V
Server Median OPM 200 users Median OPM 500 users
Optimized SQL Server 2008 DVD Store 760 1,878
Degraded SQL Server 2008 DVD Store 758 1,868
Figure 3: Median DVD Store OPM results the second Dell PowerEdge R710 server and the Dell PowerVault MD3000i achieved during
our measurement period under load for both optimized and degraded system states. Higher numbers are better.
LoadGen results
LoadGen results show the response time for the first Dell PowerEdge R710 server and Dell PowerVault MD3000i
configuration, which is the time in milliseconds it took to complete a request. Because the object of the server is to
handle as many requests as possible, lower scores, which show a server able to handle more requests in a given
time, are better.
For testing, we used 200 and 500 Microsoft Exchange mailboxes; both the server and the Dell PowerVault
MD3000i could exceed acceptable performance while performing other tasks. Figure 4 shows the average
response time for the server. Each result is the median of three runs.
Server
Median 95th percentile weighted
average for 200 users (lower is better)
Median 95th percentile weighted
average for 500 users (lower is better)
Optimized Dell
PowerEdge R710
151.20 517.45
Degraded Dell
PowerEdge R710
143.85 302.12
Figure 4: Median LoadGen response time results the first Dell PowerEdge R710 server and the Dell PowerVault MD3000i achieved
during our measurement period while under load for both optimized and degraded system states. Lower response time is better. It
is interesting to note that, with our test load, the Exchange latency was consistently better in the degraded state on all test runs.
Because the results are in response time for each task, we created a weighted average to calculate a final score.
To create a weighted average, we used a similar procedure as the one that Microsoft Exchange Server 2003
Load Simulator (LoadSim) uses. (Note: LoadSim is an older tool for testing Exchange Mail Servers; Microsoft has
replaced it with LoadGen.) Figure 5 shows the weighted average we used for testing. For an acceptable score,
the average 95th percentile latency must be lower than 1,000 ms (1 second) in all virtual machine LoadGen tests.
Task name Weight Task name Weight
BrowseCalendar 1 BrowseContacts 0
Logon 0 CreateContact 1
Logoff 0 BrowseTasks 0
SendMail 1 CreateTask 1
ReadAndProcessMessages 52 EditRules 0
MoveMail 0 DownloadOab 0
DeleteMail 0 EditSmartFolders 0
ExportMail 0 SynchronizeFolders 0
CreateFolder 0 Search 0
PublicFolderPost 0 InitializeMailbox 0
BrowsePublicFolder 0 UserInit 0
PostFreeBusy 2 UserTerm 0
RequestMeeting 1 ModuleInit 0
MakeAppointment 1 ModuleTerm 0
Total 60
Figure 5: Weighted average for LoadGen tasks.
(18 Seiten)
Manymanuals.com
Manymanuals.de
Manymanuals.fr
Manymanuals.it
Manymanuals.pl
Manymanuals.cz
Manymanuals.es
Manymanuals-pt.com
Kommentare zu diesen Handbüchern