Compaq 108164-003 - ProLiant - 800 White Paper
Disk subsystem performance and scalability
Hide thumbs
Also See for 108164-003 - ProLiant - 800:
- Integration notes (52 pages) ,
- Technical manual (38 pages) ,
- White paper (26 pages)
Advertisement
Quick Links
[September 1997]
Prepared By
Microsoft Windows NT
Integration Team
Compaq Computer
Corporation
C
O N T E N T S
Disk Subsystem
Overview ..................... 3
Disk-Related
Performance
Characteristics............ 4
Like Drive
Scalability ................. 17
Like Capacity
Scalability ................. 19
Disk Controller
Scalability ................. 24
Performance
Measurement Tools ... 27
Preventing Data
Loss while
Maintaining
Performance.............. 28
Disk Subsystem
Summary of
Findings .................... 29
ECG025.0997
W
HITE
. .
. .
Disk Subsystem Performance and
. .
. .
. .
Scalability
. .
. .
. .
. .
In today's networking environments, the disk subsystem is a key element in determining
. .
. .
overall system performance. The goal of this paper is to provide informative test results
. .
and performance-related information for various disk subsystems, to assist systems
. .
. .
engineers and network administrators in making decisions on disk subsystem installation,
. .
. .
optimization, and configuration.
. .
. .
This white paper also provides information on using Fault Tolerance to prevent data loss,
. .
. .
while maintaining system performance. Finally, this paper provides a section discussing
. .
the advantages and disadvantages of RAID technology.
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
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:
1
P
APER
CompaqNT@compaq.com
Advertisement
Related Manuals for Compaq 108164-003 - ProLiant - 800
Summary of Contents for Compaq 108164-003 - ProLiant - 800
- Page 1 Prepared By Microsoft Windows NT Scalability Integration Team Compaq Computer In today's networking environments, the disk subsystem is a key element in determining Corporation overall system performance. The goal of this paper is to provide informative test results and performance-related information for various disk subsystems, to assist systems...
- Page 2 This test is not a determination of product quality or correctness, nor does it ensure compliance with any federal, state or local requirements. Compaq does not warrant products other than its own strictly as stated in Compaq product warranties.
-
Page 3: Test Environment
This document provides disk subsystem recommendations, based on testing in the Integration Test Lab of hardware and software products from Compaq and other vendors. The test environment that Compaq selected might not be the same as your environment. Because each environment has different and unique characteristics, our results might be different than the results you obtain in your test environment. - Page 4 Disk Subsystem Testing Environment Environment Equipment Used Server Software Configuration Microsoft Windows NT Server version 4.0 Service Pack Compaq Support Software Diskette 1.20A Client Configuration Compaq Deskpro 575 Netelligent 10/100 TX PCI UTP Controller and MS-DOS NetBench 5.0 Test Configuration...
- Page 5 HITE APER (cont.) Table 2: (cont.) Disk Performance Measurement Terms Terms Description Latency The time required for the disk to spin one complete revolution. Average Latency The time required for the disk to spin half a revolution. Average Access Time The average length of time it takes the disk to seek to the required track plus the amount of time it takes for the disk to spin the data under the head.
- Page 6 HITE APER (cont.) One method to decrease seek time is to distribute data across multiple drives. For instance, the initial configuration in Figure 1 shows a single disk containing data. The new configuration reflects the data being striped across multiple disks. This method reduces seek time because the data is spread evenly across two drives instead of one, thus the disk head has less distance to travel.
-
Page 7: Average Latency
HITE APER (cont.) Average Latency Manufacturers have built and continue to build hard disks that spin at designated rates. In the early years of the personal computer (PC) industry, hard disks on the market could spin at approximately 3600 RPMs. As the market demand for better system performance increased, disk manufacturers responded by supplying faster spin rates for hard disks. - Page 8 HITE APER (cont.) However, be aware that the average latency time might not always decrease when adding more drives to your system. For example, in Figure 2 - Example 2, the new configuration shows that the amount of time it takes to retrieve the data from sector B is actually longer than the initial configuration.
-
Page 9: Transfer Rates
HITE APER (cont.) Average Access Time Average access time is simply described as average seek time plus latency. What this equates to is the amount of time the disk has to seek to find the data plus the time it takes for the disk to spin under the head. - Page 10 HITE APER (cont.) In order to share information, all of the disk subsystem components have to communicate with each other, as shown in Figure 4. The disk subsystem components communicate with each other using hardware interfaces such as Small Computer System Interface (SCSI) channels and Peripheral Component Interconnect (PCI) buses.
- Page 11 HITE APER (cont.) Disk Transfer Rates Hardware manufacturers calculate and define disk transfer rates as being the theoretical threshold for transferring data from the disk to the computer. For example, if you were to place one drive with an average transfer rate of 5 MB/s (see in Figure 4) in a system, theoretically it would take four disks to saturate a SCSI channel with a transfer rate of 20 MB/s (see...
- Page 12 Disk controllers can be a common cause of disk subsystem bottlenecks. For example, if a disk subsystem contains a Compaq Wide-Ultra SCSI Controller transferring data up to 40 MB/s, ideally it would take three controllers to saturate the PCI Bus, which transfers data at the rate of 133 MB/s.
- Page 13 To illustrate the point, it would take a minimum of three Compaq Wide-Ultra SCSI Controllers running at their maximum sustained transfer rate of 40 MB/s each to maintain throughput on the PCI Bus.
- Page 14 HITE APER (cont.) For instance, let’s say you request data stored on your disk drive (refer to Figure 4 for reference). The system first tries to complete the READ request by retrieving the data from the file system cache (memory). If it is not there, the system has to retrieve the data from the hard disk.
- Page 15 HITE APER (cont.) Table 7 lists the Host bus transfer rates for the following Compaq servers: Table 7: Host Bus (Memory) Transfer Rates Server Name Transfer Rate ProLiant 5000, 6000, 6500 and 7000 540 MB/s ProLiant 1500, 2500 and 4500 267 MB/s The last example discussed how READ performance is increased.
- Page 16 HITE APER (cont.) Concurrency Concurrency is the process of eliminating the wait time involved to retrieve and return requested data. It takes place when multiple slow devices (e.g., disk drives) place I/O requests on a single faster device (e.g., SCSI bus). As shown in Figure 6, a request for data comes across the SCSI Bus asking the disk drive to retrieve some information.
- Page 17 HITE APER (cont.) Concurrency is very effective in a multi-drive environment because, while one drive is retrieving data, another request can be coming across the SCSI bus as shown in Figure 7. When using multiple drives, each drive can send data across the SCSI bus as soon as it is available.
- Page 18 “When does the cost of adding more drives out weigh the performance gain?”. To answer this question, Compaq tested controllers using the same RAID configuration and added drives, then measured the system performance effects. Let’s now view those results and understand the effects of drive scaling.
- Page 19 Using smaller size drives limits your maximum capacity per have more drives processing disk requests. controller. For example, the Compaq SMART-2 Array Controller supports up to 14 drives. By using fourteen 2GB drives, your data capacity equals 28GB. By using fourteen 4GB drives, your data increases to 56GB.
- Page 20 HITE APER (cont.) Like Capacity Scaling Since like capacity scaling can affect your system, it is important to understand the impact it might have on system performance. To be able to determine this information, we tested like capacity scalability by maintaining the same total disk capacity for each test (8GB or 24GB) and added different quantities of drives to a single disk controller.
- Page 21 HITE APER (cont.) In Figure 10, our tests show that if you require 24 Gigabytes of storage capacity the performance gain of 33% is in using twelve 2GB disks instead of six 4GB disks. With concurrency taking place by using multiple lower capacity drives (twelve 2GB drives), more requests are being processed;...
- Page 22 HITE APER (cont.) Like Capacity Scaling (RAID 5 - 1 Controller) 14,000,000 7+1 x 2GB 12,000,000 3+1 x 4GB 10,000,000 8,000,000 6,000,000 4,000,000 2,000,000 Number of Clients Figure 11: Like Capacity Scaling in a RAID 5 Environment. The performance increase when using six 4GB drives and two 12GB drives revealed a 28% gain as shown in Figure 12.
- Page 23 Using smaller size drives limits your maximum capacity per performance (more concurrency). controller. For example, the Compaq SMART-2 Array Controller supports up to 14 drives. By using fourteen 2GB drives, your data capacity equals 28GB. By using fourteen 4GB drives, your data increase to 56GB.
- Page 24 HITE APER (cont.) Summary of Findings – Like Capacity Scaling Our test results conclude that by doubling the number of drives in a system, regardless of the data storage capacity requirements and the fault tolerance used, we consistently received an improvement in performance. However, the performance increase lessened as we added more and more drives to our system.
- Page 25 HITE APER (cont.) Multiple Disk Controller Test Results In our multiple disk controller tests we found that concurrency coupled with adding disk controllers to an environment increases system performance. For instance, Figure 13 displays a comparison between two tests in a RAID 5 environment. First, we tested twelve 4GB drives (11 data, 1 parity) using one controller.
- Page 26 HITE APER (cont.) Disk Controller Scaling (RAID 0) 20,000,000 12 x 4GB, 1 Ctr 12 x 4GB, 2 Ctr 18,000,000 16,000,000 14,000,000 12,000,000 10,000,000 8,000,000 6,000,000 4,000,000 2,000,000 Number of Clients Figure 14: Disk Controller Scaling in a RAID 0 Environment. Table 10 lists the advantages and disadvantages of disk controller scaling so that you may weigh each and decide what is right for your environment.
- Page 27 Windows NT on Compaq server hardware. These tools are now available on the Compaq Resource Paq for Microsoft Windows NT. To obtain a copy of the Compaq Resource Paq, go to the Compaq Microsoft Frontline Partnership page located on the web at: \\www.compaq.com\solutions\frontline...
-
Page 28: Fault Tolerance
Compaq Resource Paq for Microsoft Windows NT. These utilities allow easy installation and removal of Objects and Object Counters for the Compaq EISA and PCI Buses, Power Supply and NetFlex-3 Controllers. Once you install these utilities, you can view the counter data collected by the drivers through the Performance Monitor Utility included with Microsoft Windows NT. - Page 29 HITE APER (cont.) Table 13: Redundant Arrays of Inexpensive Disks Levels RAID Level Description RAID 0 (No Fault Tolerance) This RAID level is not a true fault tolerance method because it does not provide data redundancy; therefore, provides no fault protection against data loss.
- Page 30 HITE APER (cont.) Disk-Related Measurement Terms Within this document we discussed disk-related performance characteristics and how these measurement terms can affect the performance of an entire disk subsystem. Understanding how a hard disk works and the measurement terms used in the industry provides insight on the possible affect(s) disks can have on the entire disk subsystem.