Compaq ProLiant 1000 I/O Performance Tuning of Compaq Servers
Compaq ProLiant 1000 Manual
View all Compaq ProLiant 1000 manuals
Add to My Manuals
Save this manual to your list of manuals |
Compaq ProLiant 1000 manual content summary:
- Compaq ProLiant 1000 | I/O Performance Tuning of Compaq Servers - Page 1
Technical Guide March 1999 ECG044.0399 Prepared by OS Integration Engineering Compaq Computer Corporation Contents Executive Summary 3 Introduction 4 I/O Performance and Recommendations 4 Network Interface 5 Peripheral Component Interconnect (PCI) Bus 6 SCSI Bus 9 Fibre Channel 12 Drive - Compaq ProLiant 1000 | I/O Performance Tuning of Compaq Servers - Page 2
in black/white and color, and Compaq PC Card Solution logo are trademarks and/or service marks of Compaq Computer Corporation. Microsoft, Windows, Windows ©1998 Compaq Computer Corporation. All rights reserved. Printed in the U.S.A. I/O Performance Tuning of Compaq Servers Technical Guide prepared - Compaq ProLiant 1000 | I/O Performance Tuning of Compaq Servers - Page 3
with a slower network. You can prevent the adverse effects of a slow network by making sure that your server is delivering maximum performance. Compaq performs extensive tests to determine the optimal configuration for different environments. The results of these tests can be invaluable tools in - Compaq ProLiant 1000 | I/O Performance Tuning of Compaq Servers - Page 4
I/O Performance Tuning of Compaq Servers 4 Introduction Perhaps the most common term when data; the bus on this server remains relatively idle. On the other hand, large, contiguous file transfers need only position the drive heads initially to begin streaming data to the client; these requests - Compaq ProLiant 1000 | I/O Performance Tuning of Compaq Servers - Page 5
I/O Performance Tuning of Compaq Servers 5 Figure 1. Boundary Value Conditions Graphical Solution In Figure 1, the same server. In some cases, clients may be requesting random, scattered, smaller files from the server; as in a web-server. Other situations may ask the server to retrieve large, - Compaq ProLiant 1000 | I/O Performance Tuning of Compaq Servers - Page 6
connect the two PCI busses to the host bus: the dual-peer PCI bus and the bridged PCI bus. P In order to maintain maximum performance, Compaq recommends that you balance the load according to the architecture in your server. Bus Balancing in ProLiant Servers Bridged PCI bus P In bridged PCI busses - Compaq ProLiant 1000 | I/O Performance Tuning of Compaq Servers - Page 7
I/O Performance Tuning of Compaq Servers 7 the bridge chip itself, but also from any synchronization or contention with the primary bus. Simply stated, I/O loads on the secondary bus are not - Compaq ProLiant 1000 | I/O Performance Tuning of Compaq Servers - Page 8
I/O Performance Tuning of Compaq Servers 8 3. Avoid putting two network controllers together in the same bus unless both busses already have a network controller installed. Note thatSince fewer devices can lower - Compaq ProLiant 1000 | I/O Performance Tuning of Compaq Servers - Page 9
importance of a balanced bus. In general, the high-speed PCI busses in Compaq servers are less likely to become saturated in environments where random, small-block . Operating at 33 MHz and transmitting 32 bits in parallel, small files common in web serving and user storage are not usually on the - Compaq ProLiant 1000 | I/O Performance Tuning of Compaq Servers - Page 10
transfer rates up to 20 MB/s. • More Devices per Bus: Wide SCSI busses support 16 devices (15 drives, plus controller) as opposed to eight with regular ( for use with Wide SCSI. Compaq has extensively tested and integrated the Wide-Ultra SCSI-3 technology in Compaq servers and storage options because - Compaq ProLiant 1000 | I/O Performance Tuning of Compaq Servers - Page 11
Servers 11 interface and because its backward compatibility provides investment protection for Compaq customers. P In general, use devices with the highest SCSI specification that your controller can support. In the case of Fibre Channel Arrays, use Wide-Ultra SCSI-3 compatible drives for maximum - Compaq ProLiant 1000 | I/O Performance Tuning of Compaq Servers - Page 12
I/O Performance Tuning of Compaq Servers 12 Table 3. Smart-2 Controller Family Model Smart-2SL Controller Maximum Number of (Busses) Drives 1 7 2 8 1 7 2 12-1" or 8-1.6" SCSI Revision Supported Wide-Ultra SCSI-3 Wide-Ultra SCSI-3 Fast-Wide SCSI-2 Wide-Ultra SCSI-3 The SMART-2 family of - Compaq ProLiant 1000 | I/O Performance Tuning of Compaq Servers - Page 13
FC bus can carry 100 MB/s, saturating the FC bus would require at least two Fibre Channel Arrays, delivering greater than 50MB/s each. Compaq Fibre Channel Arrays are dedicated storage enclosures with integrated SMART-2 technology, each array can deliver a theoretical maximum of 80 MB/s. Here again - Compaq ProLiant 1000 | I/O Performance Tuning of Compaq Servers - Page 14
be divided evenly if dividing the drives between the SCSI busses will require mixing of SCSI revisions or placing a heavier load on one bus. Compaq storage on Fibre Channel is the Fibre Channel Array. These arrays consist of a chassis with a Fibre Channel adapter and bays for eight 1-1/2 inch drives - Compaq ProLiant 1000 | I/O Performance Tuning of Compaq Servers - Page 15
I/O Performance Tuning of Compaq Servers 15 All of these delays are associated with positioning the drive head above the requested data. The delay caused by the swinging of the - Compaq ProLiant 1000 | I/O Performance Tuning of Compaq Servers - Page 16
I/O Performance Tuning of Compaq Servers 16 service. To a client, distributing the workload means that any individual request will be filled more quickly. By spanning the data and workload to multiple devices, it - Compaq ProLiant 1000 | I/O Performance Tuning of Compaq Servers - Page 17
cache. • On-the-fly failover and rebuild without user intervention. • Drive monitoring and pre-fail warning. • Easy configuration and management. P Compaq highly recommends using hardware driven RAID systems for improved performance and reliability. Scaling The common thread in both hardware and - Compaq ProLiant 1000 | I/O Performance Tuning of Compaq Servers - Page 18
Servers 18 As mentioned, spanning your data across more drives gives a case of increased performance with diminishing returns. Compaq has found that adding drives to an array provides greater throughput regardless of storage capacity and fault tolerance level. This is true for both like - Compaq ProLiant 1000 | I/O Performance Tuning of Compaq Servers - Page 19
Performance Tuning of Compaq Servers 19 requested data is retained in high-speed RAM after the request has been filled. If this file is accessed frequently . While many modern operating systems are "auto-tuning," manual adjustment of the server environment variables and registry entries may - Compaq ProLiant 1000 | I/O Performance Tuning of Compaq Servers - Page 20
server PCI bus, Compaq NetFlex 3 network interface, and Compaq 32-Bit SCSI-2 controllers. Additionally, the Compaq Resource Paq for Microsoft Windows NT contains a utility called the Performance Monitor Analysis tool. The data in the log files created by Performance Monitor are not readily viewable - Compaq ProLiant 1000 | I/O Performance Tuning of Compaq Servers - Page 21
I/O Performance Tuning of Compaq Servers 21 Novell NetWare For the newer NetWare 5, as well as NetWare 4, MONITOR.NLM provides real-time measurements of OS performance parameters. While MONITOR.NLM, the NetWare Console Monitor, does not provide for logging data to an external file, the on-screen, - Compaq ProLiant 1000 | I/O Performance Tuning of Compaq Servers - Page 22
administrative functions like manually testing the LEDs of data. If sar's output files are too obtuse, you may -dependent problems and recommending instructions for your environment. As part of the ASE training courses, Compaq offers in-depth classes on Integration and Tuning for each of the supported - Compaq ProLiant 1000 | I/O Performance Tuning of Compaq Servers - Page 23
were collected using ZD's ServerBench 4.02, ZD's NetBench 5.01, and Intel's Iometer, rev 1998.10.08. Tests were run on a variety of Compaq servers, including ProLiant 1850R, ProLiant 6500, and ProLiant 5000. The results shown here are not intended as competitive benchmarks; rather the information - Compaq ProLiant 1000 | I/O Performance Tuning of Compaq Servers - Page 24
I/O Performance Tuning of Compaq Servers 24 Simulated Application: Web Server (8 KB, 100% Read, 0% Write, 0% Sequential, 100% Random) 900 800 700 600 500 400 300 200 100 0 RAID 0 RAID 1 RAID 5 - Compaq ProLiant 1000 | I/O Performance Tuning of Compaq Servers - Page 25
I/O Performance Tuning of Compaq Servers 25 Simulated Application: Maximum MB/sec (1 MB, 100% Read, 0% Write, 100% Sequential, 0% Random) 45 40 35 30 25 20 15 10 5 0 RAID 0 RAID 1 RAID 5 - Compaq ProLiant 1000 | I/O Performance Tuning of Compaq Servers - Page 26
I/O Performance Tuning of Compaq Servers 26 Simulated Workload: OLTP (2 KB, 67% Read, 33% Write, 0% Sequential, 100% Random) 3000 2500 2000 1500 1000 500 0 1 2 3 4 5 6 Number of Fibre Channel Arrays 5 Meter - Compaq ProLiant 1000 | I/O Performance Tuning of Compaq Servers - Page 27
I/O Performance Tuning of Compaq Servers 27 Simulated Workload: Decision Support (25% Access, 8 KB, 0% Read, 100% Write, 100% Sequential, Cables Figure 14. Relative performance of FC-AL systems for decision support applications. These results were obtained using Iometer version 1998.10.08, - Compaq ProLiant 1000 | I/O Performance Tuning of Compaq Servers - Page 28
I/O Performance Tuning of Compaq Servers 28 IO/sec vs. Number of Fibre Channel Arrays 4500 4000 3500 3000 2500 2000 1500 1000 500 0 0 1 2 3 4 5 6 7 8 9 10 11 Number of Fibre Channel Arrays 2 KB OLTP WS Mix Web Server Decision Support Figure 16. Relative sustained I/O for each number of FC arrays - Compaq ProLiant 1000 | I/O Performance Tuning of Compaq Servers - Page 29
Servers 29 I/Os per second Sequential Read: 2% Sequential Write: 1% 1400 Composite Performance Random Read: 58% Random Write: 39% 1200 1000 Representative Compaq 7,200-rpm drive Compaq 10k-II drive 800 600 400 200 With larger block sizes, the SCSI bus will saturate, limiting the I/O rate - Compaq ProLiant 1000 | I/O Performance Tuning of Compaq Servers - Page 30
I/O Performance Tuning of Compaq Servers 30 60 Read Latency - 7,200-rpm Write Latency - 7,200-rpm 50 Read Latency - 10k-II Write Latency - 10k-II 40 Latency (ms) 30 20 - Compaq ProLiant 1000 | I/O Performance Tuning of Compaq Servers - Page 31
I/O Performance Tuning of Compaq Servers 31 Throughput (transactions per second) 600 7 x 9.1 GB 7,200-rpm RAID 5 500 7 x 9.1 GB 10k-II RAID 5 400 300 200 100 0 1 2 4 6 8 12 16 20 24 28 - Compaq ProLiant 1000 | I/O Performance Tuning of Compaq Servers - Page 32
I/O Performance Tuning of Compaq Servers 32 Like Capacity Scaling Figures 23 through 27 show, again, the sustained performance of distributing loads to more drives regardless of the RAID level. - Compaq ProLiant 1000 | I/O Performance Tuning of Compaq Servers - Page 33
I/O Performance Tuning of Compaq Servers 33 Server Throughput (Bytes/sec) Like Capacity Scaling (RAID 5 - 1 Controller) 14,000,000 12,000,000 10,000,000 8,000,000 6,000,000 4,000, - Compaq ProLiant 1000 | I/O Performance Tuning of Compaq Servers - Page 34
I/O Performance Tuning of Compaq Servers 34 Drive Controller Scaling Figures 27 and 28 illustrate the performance gains of distributing parity calculations by adding RAID array controllers. In Figure 29, - Compaq ProLiant 1000 | I/O Performance Tuning of Compaq Servers - Page 35
determine which balance of fault tolerance and speed are best for your environment. 1000 900 800 700 600 500 400 300 200 100 0 Compaq Smart-2DH SCSI Controller Simulated Application: OLTP (2 KB, 67% Read, 33% Write, 0% Sequential, 100% Random) RAID 0 RAID 1 RAID 5 RAID Level RAID 5 No Cache - Compaq ProLiant 1000 | I/O Performance Tuning of Compaq Servers - Page 36
. These results were obtained using Iometer version 1998.10.08, Copyright 1998 by Intel Corporation. Intel does not endorse any Iometer results. Compaq Smart-2DH SCSI Controller Simulated Application: Maximum MB/sec (1 MB, 100% Read, 0% Write, 100% Sequential, 0% Random) 45 40 35 30 25 20 - Compaq ProLiant 1000 | I/O Performance Tuning of Compaq Servers - Page 37
I/O Performance Tuning of Compaq Servers 37 30000 25000 (256 KB Block Size, 100% Read, 100% Sequential) 26053 25874 KB/sec 20000 15000 10000 5000 10480 7656 17000 SMART-2SL - Compaq ProLiant 1000 | I/O Performance Tuning of Compaq Servers - Page 38
I/O Performance Tuning of Compaq Servers 38 IO/s for a Common Database Profile (8KB Block Size, 67% Read, 100% Random) 450 400 377 350 335 300 288 IO/sec 250 200
Technical Guide
Help us improve our technical communication. Let us know what you think
about the technical information in this document. Your feedback is valuable
and will help us structure future communications. Please send your
comments to:
March 1999
ECG044.0399
Prepared by OS
Integration
Engineering
Compaq Computer Corporation
Contents
Executive Summary
...................
3
Introduction
................................
4
I/O Performance and
Recommendations
.....................
4
Network Interface
......................
5
Peripheral Component
Interconnect (PCI) Bus
..............
6
SCSI Bus
..................................
9
Fibre Channel
.........................
12
Drive Technology
....................
14
System Memory
......................
18
OS Performance Monitoring
Tools
.........................................
19
Microsoft Windows NT
............
20
Novell NetWare
.......................
21
SCO UnixWare and
OpenServer
............................
21
Conclusion
..............................
22
Appendix - Performance
Tuning Data
..............................
23
Fibre Channel Bus
..................
23
Drive Technology
....................
28
RAID Level Performance
.........
35
Hardware RAID vs. Software
RAID
......................................
36
I/O Performance Tuning of
Compaq Servers
Abstract:
Tuning the disk system and I/O paths can be key to
achieving maximum performance from your server. This paper
begins with definitions, guidelines, and recommendations for I/O
performance tuning. The last section of this paper provides actual
performance data to reinforce recommendations. In general, this
paper deals with universal I/O concerns and applies to a wide range
of server applications. The major areas examined in this paper are
•
disk systems
•
NICs
•
memory
•
system configuration
•
performance monitoring tools.