Benchmark Tests - Revelation Windows 2000 Service 2.1 on Windows NT/2000 Servers running 16-bit OpenInsight (Performance)
Created at 15 MAR 2002 03:22PM
Table of Contents
Notes on Revelation's Network Products Benchmark Tests[#NPB1]
Standard Test Bank[#Fig1]
NT 4.0[#NPB4]
Windows 2000[#NPB5]
[<a name="NPB1">]Notes on Revelation's Network Products Benchmark Tests[</a>]
Revelation is pleased to publish the results of its internal network testing procedures. The purpose of these test is to:
ascertain the quality of the Revelation Network Products by performing tests that simulate extreme processing conditions.
determine the relative performance of the Revelation Windows 2000 Service against the performance the Revelation Windows 2000 Service 1.5.
Methodology
Revelation has developed a standard test bank[#Fig1]. Under each individual network heading, there is a table that summarizes the elements of the testing environment, including network, PC types, OpenInsight version used, services, and number of loops performed. Following is the results in raw data.
Conclusions
When running Named Pipes only, there is little performance difference between running the Windows 2000 Service version 1.5 and 2.1.
When running the 2.1 Service using Named Pipes and Named Pipes to TCP/IP, the speed increases can be dramatic, as much as 60% faster.
We encourage you to use these tests in your environments and compare results to those that we obtained. Please note that performance is only one of the benefits offered by the Windows 2000 Service. Most users have experienced an incredible increase in network stability through the elimination of network errors (GFEs). Please refer to the Product Info section of our web site for complete information on the value of Revelation s Network Product portfolio.
[<a name="Fig1"]Standard Test Bank[</a>]
Test | Test Code | Description |
Lock/Write/Delete/Unlock | LWDU | 1. Open LH table. Each iteration of the loop corresponds to a record key. 2. Lock record at record key iteration. If lock fails, wait 1 second, then retry lock. 3. Write record at record key iteration. 4. Delete record at record key iteration. 5. Unlock record at record key iteration. |
Lock/Write/Read/Unlock | LWRU | 1. Open LH table. Each iteration of the loop corresponds to a record key. 2. Lock record at record key iteration. If lock fails, wait 1 second, then retry lock. 3. Write record at record key iteration. 4. Read record at record key iteration. 5. Unlock record at record key iteration. |
Lock/Write/Unlock | LWU | 1. Open LH table. Each iteration of the loop corresponds to a record key. 2. Lock record at record key iteration. If lock fails, wait 1 second, then retry lock. 3. Write record at record key iteration. 4. Unlock record at record key iteration. |
Read | R | 1. Open LH table. Each iteration of the loop corresponds to a record key. 2. Read record from table iteration. |
Select/ReadNext | SRN | 1. Open and select LH table. Each iteration of the loop corresponds to a record key. 2. ReadNext key. |
Select/ReadNext/Read | SRNR | 1. Open and select LH table. Each iteration of the loop corresponds to a record key. 2. ReadNext key. 3. Read record from table at key. |
[<a name="NPB4">] NT 4.0 [
Network | Server Type | PC Types | OpenInsightVersion | Services | Loops |
NT 4.0 SP1 | PII 266, 80MB | Win 98SE, PI 166 64MB Win NT4.0 SP5 PII 266, 64MB Win2K no SP PII 300 128MB Win XP AMD 500 256MB | 3.7.4 | 2.1 Service with Named Pipes Only 2.1 Service with Named Pipes → TCP/IP 2.1 Service with TCP/IP Only 1.5 Service with Named Pipes Only (Baseline) | 5,000 10,000 20,000 |
Benchmark Tests
2.1 Service with Named Pipes Only on Server Portion
2.1 Service with Named Pipes → TCP/IP on Server Portion
2.1 Service with TCP/IP Only on Server Portion
[<a name="NPB5">] Windows 2000 [
Network | Server Type | PC Types | OpenInsightVersion | Services | Loops |
Windows 2000 Active Directory | PII 300, 256MB | Win 98SE, PI 166 64MB Win NT4.0 SP5 PII 266, 64MB Win2K no SP PII 300 128MB Win XP AMD 500 256MB | 3.7.4 | 2.1 Service with Named Pipes Only 2.1 Service with Named Pipes → TCP/IP 2.1 Service with TCP/IP Only 1.5 Service with Named Pipes Only (Baseline) | 5,000 10,000 20,000 |
Benchmark Tests
2.1 Service with Named Pipes Only on Server Portion
2.1 Service with Named Pipes → TCP/IP on Server Portion
2.1 Service with TCP/IP Only on Server Portion