Exchange 2000 MAPI Messaging Benchmark (MMB2) Performance Result
Hardware: / Dell™ PowerEdge™ 2650Software: / Exchange 2000 Server Enterprise Edition, SP3
Test Profile: / MAPI Messaging Benchmark
Test Date: 23 Oct., 2003
The new[this is no longer new if it is more than 6 months old. I thought MS had recently come out with MMB3; if that is the case, why aren’t we using that new benchmark?] MAPI Messaging Benchmark (MMB2) measures throughput in terms of a specific profile of user actions, executed over an 8-hour working day.
This benchmark is different from the ‘Medium User’ setting that was used with Exchange 5.5 in that the rate of client requests is significantly greater for this MMB2 profile.
Results should be interpreted as a benchmark for messaging throughput and should not be confused with deployment recommendations. Factors such as backup/restore, topology and other issues should be considered when planning a deployment. For information on how MMB2 results differ from deployment and configuration information, refer to the note below on Benchmark vs. Production Configuration.
Summary of Results
The Dell PowerEdge 2650 demonstrates the ability to support a high number of MMB2s in a direct attached SCSI configuration running Exchange 2000 Server Enterprise Edition on Windows® 2000 Advanced Server. The PE2650 supported 11,600 MMB2s. Microsoft’s LoadSim tool was used to simulate 11,600 Exchange (MAPI) e-mail users running on a Dell PowerEdge 16502650 Rack Server equipped with two 3.06GHz Xeon®™ processors and 4GB of memory. The canonical "MMB2" profile was used, which represents a typical corporate e-mail user. During this period the PowerEdge 2650 provided a weighted 95th percentile response time of 137 ms for 11,600 users, with an average CPU utilization of 83.52%.
Results are based on 4 hours of steady state running.
Benchmark vs. Production Configuration Disclosure Note
This test measures the messaging throughput of a single server, single site topology. Its purpose is to measure the maximum throughput of a Microsoft® Exchange Server on this hardware configuration. This can provide a benchmark for comparing hardware and/or software products, but cannot be used as a deployment guide for production environments. For deployment specific information, please contact a Microsoft or Dell representative.
The MMB2 benchmark does not account for:
· Usage profiles not matching that of the Load Simulator MAPI Medium profile
· Per user storage, and per server backup requirements
· Fault tolerance requirements
· Workloads other than MAPI private folder access. This includes Public Folder, NNTP, POP3 and other email interfaces
· Multiple Exchange Server deployments, where additional resources are required to forward mail intra-site
· Connectors, Links and replication to remote Exchange sites
Test Results
SummarySupported Benchmark Load / 11 6,6200 MMB2s
Benchmark Profile / MAPI Messaging Benchmark 2 (MMB2)
Protocol / Exchange MAPI
Length of Steady State / 4 Hours
Length of Test / 8 Hours
Unless otherwise noted, values listed below are averages over
entire 4-hour steady state period.
Transactions in total
Total Messages Submitted / 156,130282,107
Total Message Recipients Delivered / 576,0121,025,973
Total Messages Sent / 156,126282,025
Ratio Message Recipients Delivered / Messages Submitted / 3.6964
Transaction Load (per hour)
Messages Submitted / hour / 38,98470,375
Message Recipients Delivered / hour / 143,950255942
Messages Sent / hour / 38,98370,355
Transaction Load (per Second)
Message Opens/Sec / 60.496113.9
Folder Opens/Sec / 25.84358.4
RPC Read Bytes/Sec / 140,808268279
RPC Write Bytes/Sec / 1,123,8932,167,057
Transaction Queues
IS Send Queue Average Length / 82.62583
Processor Utilization
System Processor Utilization (%) / 92.10583.525
System Processor Queue Length / 10.7259
System Context Switches/Sec / 7059.30514,087.816
Process % CPU Time - Store / 146.784282.021
Process % CPU Time - Inetinfo / 7.74113.573
Exchange 2000 server is also domain controller? (yes/no) / Yes
Process % CPU Time – LSASS (on domain controller) / 12.54120.239
Memory Utilization
Available Bytes / 2.3111.74 GB
Pages/Sec / 0.9552.956
Process Working Set Bytes - Store / 1.21 255 GB
Process Virtual Bytes - Store / 1.962.058 GB
Logical Drive Utilization
IS Database Disk Reads/Sec / (G) 243.121+ (H) 242550+ (I) 246.390+ (J) 244.651= 2,142.733976.712
IS Database Disk Writes/Sec / (G) 138.667+ (H) 132.519+ (I) 137.037+ (J) 142.603= 1,281.605550.826
IS Database Average Disk Queue Length / (G) 11.186+ (H) 10.957+ (I) 12.297+ (J) 11.749=
11.54710.616
IS Log Disk Reads/Sec / (L) 0.0+ (N) 0.0 = 0.00.001
IS Log Disk Writes/Sec / (L) 177.626+ (N) 176.849= 1,223.375354.475
IS Log Average Disk Queue Length / (L) 0.108+ (N) 0.107 = 0.10750.211
Descriptive Terms
Messages Submitted
Submit calls made by clients. This equates to total messages sent by users.
Messages Sent
Messages that the Store sends to the categorizer in Inetinfo (SMTP Service in particular). [1]
Message Recipients Delivered
Separate mailboxes that messages have been delivered to.
Message Opens/Sec
Messages accessed for reading per second.
Folder Opens/Sec
Folders opened for browsing per second.
RPC Read Bytes/Sec
Bytes read from clients, sent via RPCs.
RPC Write Bytes/Sec
Bytes written to clients, sent via RPCs.
IS Send Queue Average Length
Send Queue Size is the number of messages in the private information store's send queue.
Response Times (Latencies)
Client Actions /95th Percentile Response Time
(in Milliseconds)
Read / 170100Send / 350230
Delete / 10060
Move / 200120
Submit / 16090
Weighted Total / 235137
Message Throughput
Summary of the MMB2 profile for an 8-hour day:
Expected / MeasuredMessages Submitted/MMB2/Day / 51 / 50.348.5
Messages Delivered/MMB2/Day / 185 / 185.6176.5
Average Recipients per Message / 3.6 / 3.6964
· No modifications were made to the default profile.
Server Configuration
Hardware / Exchange Server / Domain Controller (if remote)Vendor / Dell
Model / PowerEdge 21650
Processor / 3.061 .4 GHz Pentium®III Xeon MP??
# Of Processors / Two
Primary Cache / 32KB
Secondary Cache / 512KB
Tertiary Cache / 1 MB
Memory / 4GB
Disk Subsystem / 56 112 18.35 4 GB disks in eight PowerVault™ 220S SCSI enclosures
14 28 18.35 4 GB disks in one two PowerVault 220S SCSI enclosures
Two Five 3618.35.7 GB internal disks
Disk Controllers / One PERC 3/Di Embedded controller with 128MB Cache and two three Mylex eXtremeRAID 2000 LSI Logic MegaRAID Elite 1650 RAID controllers with 64MB 32MB Cache each
Other Hardware
Hardware Tunings / One Intel PRO/1000 XT Embedded Broadcom Gigabit Adapter. Flow Control set to Auto, IP and TCP Checksum Offload and Transmit enabled. Write Back Cache enabled on controllers and disk drives.
Comments
Mail Software
Vendor / Microsoft Corporation / N/A
Mail Server / Exchange Server 2000 / N/A
Build\Release Version / Enterprise Edition with SP3 / N/A
Additional Software Tuning / /3GB switch added to boot.ini / N/A
Mailbox Store database maintenance intervals set to off via “Custom” option.
msExchESEParamLogBuffers = 500
HeapDeCommitFreeBlockThreshold = 40,000Default screen saver disabled.
OS Software
Operating System\Version / Windows 2000 Advanced Server
Service Pack\Patch Info / SP3
File System Type / NTFS, Dynamic Volume for Logs and Information Stores
Other Software
Network
Type of Network / Ethernet
Network Speed / 1Gbps Full Duplex
MSL (sec)
Time-Wait (sec)
Load Generator Configuration
# of Load Generators (LG) / 16Total # of LG processes / 11,6006,200
Simulated Users/Process / 165 clients were configured with 725388 users each, one client configured with 380 users
Model / 16 Dell PowerApp.web 100s
Processor / 1.0 1GHz Pentium III
# of Processors / 11
Memory / 512MB
Network Controller / Intel 82559 Fast Ethernet LAN on Motherboard
Operating System / Windows 2000 Professional, SP3
[1] All messages -- even MAPI messages – are sent to the categorizer, as this replaces the MTA for all but communication via X.400 with an Exchange 5.5 server.