Compaq 850R - ProLiant - 32 MB RAM User manual


November 1998
ECG082/0898
Prepared by Messaging and
Collaboration Solutions Business
Unit
Compaq Computer Corporation
Contents
Compaq ProLiant 7000............... 3
Exchange Server
Performance Test Results.......... 3
What the Benchmarks Don’t
Tell You ..................................... 4
MAPI Messaging Benchmark
(MMB) - LoadSim Medium
User Redefined ......................... 4
MMB Transaction Load............. 4
Message Mix Description.......... 5
Related Documents .................... 9


 !!!
""#
Abstract: Compaq achieved record-breaking Exchange Server
scalability by achieving 15,000 MMB on a ProLiant 7000 server
equipped with four 400-MHz Intel Pentium

II Xeon

processors.
Using Microsoft’s Load Simulation utility, the ProLiant 7000 with
four 400-MHz processors was tested at Compaq’s Microsoft
Competency Center in Redmond, Washington. The favorable results
of this testing demonstrate the superior performance capabilities of
four-processor servers based on the Intel Pentium

II Xeon

processor as compared to eight-processor servers based on the Intel
Pentium

Pro available from competitors.
Compaq and Microsoft have worked to optimize Microsoft Exchange
Server performance on Compaq server products to provide an
optimal balance between performance, availability, manageability,
and cost. Compaq not only provides world-class server platforms,
but also the experience necessary for the successful deployment of
messaging and collaborative applications

 ! "
ECG082/0898
Notice
The information in this publication is subject to change without notice and is provided “AS IS” WITHOUT
WARRANTY OF ANY KIND. THE ENTIRE RISK ARISING OUT OF THE USE OF THIS
INFORMATION REMAINS WITH RECIPIENT. IN NO EVENT SHALL COMPAQ BE LIABLE FOR
ANY DIRECT, CONSEQUENTIAL, INCIDENTAL, SPECIAL, PUNITIVE OR OTHER DAMAGES
WHATSOEVER (INCLUDING WITHOUT LIMITATION, DAMAGES FOR LOSS OF BUSINESS
PROFITS, BUSINESS INTERRUPTION OR LOSS OF BUSINESS INFORMATION), EVEN IF
COMPAQ HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH DAMAGES.
The limited warranties for Compaq products are exclusively set forth in the documentation accompanying
such products. Nothing herein should be construed as constituting a further or additional warranty.
This publication does not constitute an endorsement of the product or products that were tested. The
configuration or configurations tested or described may or may not be the only available solution. This test
is not a determination or product quality or correctness, nor does it ensure compliance with any federal
state or local requirements.
Product names mentioned herein may be trademarks and/or registered trademarks of their respective
companies.
Compaq, Contura, Deskpro, Fastart, Compaq Insight Manager, LTE, PageMarq, Systempro, Systempro/LT,
ProLiant, TwinTray, ROMPaq, LicensePaq, QVision, SLT, ProLinea, SmartStart, NetFlex, DirectPlus,
QuickFind, RemotePaq, BackPaq, TechPaq, SpeedPaq, QuickBack, PaqFax, Presario, SilentCool,
CompaqCare (design), Aero, SmartStation, MiniStation, and PaqRap, registered United States Patent and
Trademark Office.
Netelligent, Armada, Cruiser, Concerto, QuickChoice, ProSignia, Systempro/XL, Net1, LTE Elite,
Vocalyst, PageMate, SoftPaq, FirstPaq, SolutionPaq, EasyPoint, EZ Help, MaxLight, MultiLock,
QuickBlank, QuickLock, UltraView, Innovate logo, Wonder Tools logo in black/white and color, and
Compaq PC Card Solution logo are trademarks and/or service marks of Compaq Computer Corporation.
Microsoft, Windows, Windows NT, Windows NT Server and Workstation, Microsoft SQL Server for
Windows NT are trademarks and/or registered trademarks of Microsoft Corporation.
Intel, Pentium, Pentium II, Xeon, and Pentium Pro are trademarks and/or registered trademarks of Intel
Corporation.
Cheyenne, ArcServe and Computer Associates are the trademarks and/or registered trademarks of
Computer Associates, Inc.
NetWare and Novell are registered trademarks and intraNetWare, NDS, and Novell Directory Services are
trademarks of Novell, Inc.
Copyright ©1998 Compaq Computer Corporation. All rights reserved. Printed in the U.S.A.
Enterprise-Class Microsoft Exchange Server Scalability on Compaq ProLiant 7000 Pentium II Xeon
Servers
Solutions Guide prepared by Messaging and Collaboration Solutions Business Unit
First Edition (November 1998)
Document Number ECG082/0898

 ! #
ECG082/0898
Compaq ProLiant 7000
The Compaq ProLiant 7000 is the ultimate standards-based server delivering the most scalable
performance and the highest levels of availability. The ProLiant 7000 supports up to four
PentiumII Xeonprocessors and 8 GB of system memory. Its architecture supports Enhanced
PCI Hot-Plug and a Triple-Peer PCI bus design (five 64-bit PCI, four 32-bit PCI, and one ISA
slot). Standard with the ProLiant 7000 is the 64-bit PCI SMART 3100ES Array Controller that
supports 3 Wide Ultra SCSI-3 channels and 64 MB of cache. Leveraging Fibre Channel, the
Compaq ProLiant 7000 can support nearly 7 TB of external storage. Combined with the latest
high-availability features and processor technology, the ProLiant 7000 is designed for the most
demanding and mission-critical applications.
Exchange Server Performance Test Results
The tests were conducted using Microsoft MAPI Messaging Benchmark (MMB). The MAPI
Messaging Benchmark (MMB) measures throughput in terms of a specific profile of user actions,
executed over an 8 hour working day. This benchmark utilizes the ‘Medium User’ setting of the
Load Simulator MAPI tool. 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.
Table 1. Performance Highlights (ProLiant 7000, (4) PentiumII Xeon400-MHz, 4 GB RAM)
MMB 15,000
Response Time (milliseconds) 343
Messages Submitted (8-hour period) 213,021
Messages Delivered (8-hour period) 260,707
Message Recipients Delivered (8-hour period) 1,179,107
Messages Sent (8-hour period) 57,315
Note: Complete Test results disclosure can be found later in this document.
Table 2. Configuration of the tested Compaq ProLiant 7000
Compaq ProLiant 7000 Tested Configuration
(4) PentiumII Xeon400-MHz – 1 MB L2 cache per processor
4 GB RAM
(1) SMART 3100ES Array Controller with 64 MB Cache (configured for 100% WB)
OS/Pagefile/Exchange DS/MTA Files/Exchange Log Files: (2) 4.3-GB Drives – RAID1
Exchange Information Store Files: (16) 4.3-GB Drives – RAID0
Compaq Netelligent (100BaseTX) network interface card (NIC)
Windows NT Server v4.0 Enterprise Edition (using /3GB BOOT.INI switch)
Exchange Server v5.5 – Enterprise Edition
Note: Complete Test results disclosure can be found later in this document.

 ! $
ECG082/0898
What the Benchmarks Don’t Tell You
It is important to understand that benchmarks such as these are designed to give Exchange Server
implementation planners baseline references for understanding the capabilities of hardware
platforms from a single vendor such as Compaq or other competing hardware vendors. When
interpreting these benchmarks, two things should be kept in mind.
First, consider whether benchmarks are performed on what can be referred to as customer-
deployable configurations. A hardware vendor may publish a result that is based on a platform or
configuration that should not be deployed in a real-world Exchange Server deployment. For
example, many vendors (including Compaq) publish results using disk subsystems configured
with RAID0. While RAID0 does provide the highest levels of disk subsystem performance, it
fails to provide any protection against data loss. Compaq recommends deploying an Exchange
Server with disk fault tolerance such as RAID1 or RAID5 for the highest levels of data
protection.
Second, most vendors, including Compaq, conduct benchmarks for Exchange Server that are
single-server in nature. Also, keep in mind that benchmarks do not account for issues such as
backup and disaster recovery or information store maintenance sizing. Whatever the issue, care
must be taken when interpreting benchmarks to ensure that they represent useful information for
your Exchange Server deployment and are based on valid simulation methodologies.
While it is significant that the ProLiant 7000 server can successfully scale to 15,000 medium
MAPI e-mail users (MMB) in a single-server benchmark exercise, Compaq recommends careful
evaluation of all issues involved in real-world Exchange Server deployments – issues such as
management, administration, and disaster recovery.
MAPI Messaging Benchmark (MMB) - LoadSim Medium
User Redefined
To distinguish clearly between throughput benchmarks and capacity planning information for
Microsoft Exchange Server, Microsoft has established the MAPI Messaging Benchmark (MMB)
based on the workload from LoadSim Medium User profile. The MAPI Messaging Benchmark
representative workload focuses on the resulting throughput and clearly communicates the profile
under test.
The workload profile has not changed from the former LoadSim Medium User, but expressed in
clearer fashion. The intent is to make sure that customers can understand the MAPI Messaging
Benchmark workload and can compare it to other platforms. In addition, the renaming of the
benchmark reinforces that the test is a measurement of messaging throughput and that additional
considerations are required in capacity planning.
MMB Transaction Load
The transaction load created by the benchmark is equivalent to the user actions over an eight-hour
day as shown in Table 3.

 ! %
ECG082/0898
Table 3. Average User Actions
User Action Actions Per Day
Check Inbox 12
Send Message 14.18
Avg. Recipients per Message 4.7
Messages Received 66.3
Read Message 81.3
Move Message 16.3
Delete Message 32.5
Update Calendar 5
30% of all mail messages have one distribution list recipient. Average size of DL 10 recipients.
(Recipients created by distribution lists are included in the summary above.) All users are logged
on prior to the benchmark as they are assumed to be using mail in a corporate setting. Mail is not
cleared from the deleted items folder during the test as this is assumed to happen when the user
logs off.
Message Mix Description
The weights used when the Load Simulator randomly selects which message to send are listed in
the Table 4.
Table 4. Message Mix Description
Message Files Body Attachment Content Description Weight
Ups1k.msg  Body as RTF 60
Ups2k.msg  Body as RTF 16
Ups4k.msg  Body as RTF 4
Ups10kat.msg  10K Body as RTF - Notepad attachment 6
Upsxlatt.msg  15K Body as RTF - Microsoft Excel spreadsheet attached 4
Upswdatt.msg  16K Body as RTF - Microsoft Word document attached 4
Upsbmobj.msg  43K Body as RTF - Bitmap attachment 2
Load Simulator
The main tool used in generating the workload the MMB benchmark was Microsoft Load
Simulator. Load Simulator is a tool for simulating a client user load on an Exchange Server. Its
purpose is to enable a single Windows NT machine called a LoadSim client to simulate multiple
Microsoft Exchange client users.
The operation of Load Simulator users is governed by a Load Simulator profile. This profile
controls factors such as how long a Load Simulator "day" is, how many e-mail messages to send
in a day’s time, how many times to open and read e-mail, whether to use distribution lists,
whether to use public folders, etc.

 ! &
ECG082/0898
Load Simulator creates a highly accurate simulation of reality. It mimics the full Microsoft
Exchange Client in many respects. First, it uses .MSG files, the same format used by the
Exchange Client. This guarantees that messages generated by Load Simulator have the same
properties as those sent by real users of the Exchange Client. Second, Load Simulator uses the
same MAPI remote procedure call (RPC) semantics as those used by the Client. Third, Load
Simulator registers MAPI change notifications in the same manner as they are registered by the
Client. Finally, Load Simulator even emulates the Microsoft Exchange Client list box cache,
which the Client uses for folder and message panes in the viewer when a user browses and selects
messages on the server. For more information on LoadSim Medium canonical profiles, refer to
the LoadSim documentation at http://www.microsoft.com/exchange/library/loadsim55x86.exe.
Table 5. TEST DISCLOSURE
LoadSim Clients Configuration
Network Topology (100Base T, Token Ring, etc.) 100 Base-TX
Number and type of clients (≤40) 2x5/133, 128 MB RAM (≤400 users each)
or better (indicates minimum configuration)
Number and type of hubs/concentrators (full
duplex, switching, etc.) Compaq Netelligent 5708 Switch and Netelligent 2624 Hubs
Number of clients/segment 20
Client CPU type and speed in percentages 2P/133-MHz Pentiumprocessors or better
Client network controller broken down by
percentages Compaq Netelligent 10/100
Client network software name and version (drivers,
protocols, redirector) Microsoft Windows NT Workstation 4.0 with SP3
TCP/IP
Size of any client network cache None
Network controller software Compaq Netelligent 10/100 driver
LoadSim version 5.5 (Build 2187)
Note: Response time measurements were taken from a LoadSim Control Client simulating 100
users configured with 96 MB RAM and a Pentium/166 CPU. The client is located on an isolated
network segment connected to a 100-Mb/s switch.
Benchmark Tuning
Table 6 shows tuning parameters for Exchange Server that varied from default settings for
purposes of the benchmark exercise. Compaq recommends using the default settings specified by
the Microsoft Exchange Performance Optimizer for production server deployments.
Table 6. Benchmark Tuning
Parameter Default Setting Benchmark Setting
IS Buffers (4K each) 524,000 375,000
Max Open Tables 32,000 200,000
IS Send Threads 4 8

 ! 
ECG082/0898
IS Deliver Threads 4 8
MTA Submit/Deliver Threads 2 8
MTA TCP Threads 2 6
MTA TCP Control Blocks 20 25
Table 7. Version Information (Server)
Name Type Version or Date
Compaq NT SSD Software/Drivers 2.09
SMART ECI Driver Software/Drivers 1.06
System Firmware ROM 7/8/98
SMART 3100ES Firmware ROM 2.72
Windows NT Server EE Operating System 4.0 SP3
Exchange Server EE Application 5.5 (Build 1960.8)
Table 8. Performance Data
15,000 MMB (Measured during test run at steady state)
Summary
Supported Benchmark Load 15,000 MMB
Weighted 95th-Percentile Score 343 milliseconds
Benchmark Profile MMB (Medium MAPI User)
Protocol Exchange MAPI
Length of Steady State 4 hours
Length of Test 8 hours
Transaction Load (hourly)
Messages Submitted 26,627.69
Message Recipients Delivered 147,388.46
Messages Sent 7,164.49
Transaction Load (per Second)
Message Opens/Sec 53.75
Folder Opens/Sec 14.05
RPC Read Bytes/Sec 47,250.31
RPC Write Bytes/Sec 383,111.15
Transaction Queues
IS Send Queue Average Length 5.50
MTA Work Queue Average Length 8.04

 ! '
ECG082/0898
Processor Utilization
System Processor Utilization (%) 83.74%
System Processor Queue Length 4.29
System Context Switches/Sec 5438
Process % CPU Time – Store 272.33%
Process % CPU Time – DS 25.38%
Process % CPU Time – MTA 15.91%
Memory Utilization
Available Bytes 1838 MB
Pages/Sec 0.015
Process Working Set Bytes – Store 1.99 GB
Process Virtual Bytes – Store 2.78 GB
Logical Drive Utilization
IS Database Disk Reads/Sec 409.40
IS Database Disk Writes/Sec 244.01
IS Database Average Disk Queue
Length 11.36
IS Log Disk Reads/Sec 0.93
IS Log Disk Writes/Sec 159.82
IS Log Average Disk Queue Length 0.34
Note: Performance results were measured using Microsoft NT Performance Monitor.
Measurements were obtained by measuring averages for the period of steady-state activity (i.e.
after 15, 000 users were successfully logged on). Tests measure the messaging throughput of a
single-server, single-site topology. For deployment-specific information contact a Microsoft or
Compaq representative. More information can be found at:
http://www.microsoft.com/exchange/support/deployment/planning/deploy.asp?A=5&B=1
Table 9. User Response Times (Latencies) from Load Simulator
Client Actions 95th-Percentile Response Time (in Milliseconds)
Read 260
Send 531
Delete 421
Move 550
Submit 441

 ! (
ECG082/0898
Related Documents
The following documents are available on the Compaq website:
Compaq and Microsoft Demonstrate Enterprise Scalability with Exchange Server 5.5,
http://www.compaq.com/support/techpubs/whitepapers/ECG00961197.html
Microsoft Exchange Server 5.5 on the Compaq ProLiant 850R,
http://www.compaq.com/support/techpubs/whitepapers/ECG0710698.html
Microsoft Exchange Server 5.5 on the Compaq ProLiant 3000,
http://www.compaq.com/support/techpubs/whitepapers/ECG0720698.html
Microsoft Exchange Server 5.5 on the Compaq ProLiant 6000 Class Servers,
http://www.compaq.com/support/techpubs/whitepapers/ECG0730698.html
Compaq Deployment and Configuration Guide: Microsoft Exchange Server on Compaq ProLiant
Servers,
http://vcmproapp02.compaq.com/
Performance of Exchange Server 4.0 on Compaq ProLiant Servers,
http://www.compaq.com/support/techpubs/whitepapers/444A0696.html
“Deschutes” Family Processor Technology,
http://www.compaq.com/support/techpubs/whitepapers/ecg0500698.html
Disk Subsystem Performance and Scalability,
http://www.compaq.com/support/techpubs/whitepapers/ECG0250997.html
Configuring Compaq RAID Technology for Database Servers,
http://www.compaq.com/support/techpubs/technotes/184206-1html
Compaq SMART Array Controller Technology,
http://www.compaq.com/support/techpubs/whitepapers/667A0697.html
Hardware vs. Software Fault Tolerance,
http://www.compaq.com/support/techpubs/whitepapers/ECG0660298.html
Compaq Pentium Pro Processor-based Servers,
http://www.compaq.com/support/techpubs/whitepapers/308A0496.html
Configuring the Compaq ProLiant 5000 Server for Peak Performance,
http://www.compaq.com/support/techpubs/whitepapers/679A0697.html
Compaq White Paper Index,
http://www.compaq.com/support/techpubs/whitepapers