BASELINES

J.D. EDWARDS RESEARCH SERIES Understanding, Comparing, and Testing OneWorld Technologies

A Performance Benchmark Report

Performance Differences between Remote Desktop Protocol (RDP) and Independent Computing Architecture (ICA)

Technology Demographic Table Product

OneWorld

Version

B7322

Platform/OS

Windows Terminal Server/Citrix MetaFrame

Industry

all

Application

all

Database Keywords

Date

? J.D. Edwards World Solutions Company. All Rights Reserved

Oracle 8.04 Citrix, Independent Computing Architecture, ICA, Remote Desktop Protocol, RDP, TSE, Windows Terminal Server Edition January 4, 1999

05/24/01

Performance Differences between Remote Desktop Protocol (RDP) and Independent Computing Architecture (ICA)

Table of Contents EXECUTIVE SUMMARY ...................................................................................................................1 RESULTS..................................................................................................................................................1 INTRODUCTION...................................................................................................................................2 RUNNING TSE A LONE AND REMOTE DESKTOP PROTOCOL (RDP)................................................2 CITRIX METAFRAME AND INDEPENDENT COMPUTING A RCHITECTURE (ICA)............................2 Additional Advantages to Using MetaFrame...............................................................................3 TEST ENVIRONMENT........................................................................................................................3 SIGN-ON TESTS.....................................................................................................................................3 SIGN -ON TEST ONE – RDP HIGH SPEED LAN CONNECTION .........................................................5 SIGN -ON TEST TWO – RDP LOW SPEED WAN CONNECTION........................................................6 SUMMARY OF SIGN -ON TESTS.............................................................................................................6 SCRIPTED APPLICATION TESTS – LAN ...................................................................................7 A DDRESS BOOK LAN TESTS................................................................................................................7 ITEM M ASTER LAN TESTS...................................................................................................................8 REQUISITION CREATION LAN TEST ....................................................................................................9 SUMMARY OF LAN TEST DATA ..........................................................................................................9 SCRIPTED APPLICATION TESTS – WAN............................................................................... 10 Address Book WAN Test............................................................................................................. 10 Item Master WAN Test................................................................................................................. 10 Requisition Creation WAN Test................................................................................................. 11 SUMMARY OF WAN TEST DATA.......................................................................................................11 CONCLUSIONS AND RECOMMENDATIONS........................................................................ 12 CONTACT ............................................................................................................................................. 12

J.D. Edwards Research Series – Baselines

Performance Differences between Remote Desktop Protocol (RDP) and Independent Computing Architecture (ICA)

Executive Summary Microsoft NT Server – Terminal Server Edition (TSE) allows users to run applications, such as OneWorld, on a remote terminal server as if those applications were running on a local desktop. The TSE operating system greatly reduces Wide Area Network (WAN) traffic by downloading only screen presentations from a terminal server, and only uploading keystrokes and mouse movements from client workstations. In September, 1998 Worldwide Advanced Technologies conducted preliminary sizing tests running OneWorld in conjunction with TSE, and demonstrated some advantages to purchasing the TSE add-on, Citrix Metaframe ? . Following our initial investigation, a customer indicated interest in further TSE/MetaFrame testing, and participated in the study reported here. In this paper, we demonstrate some performance advantages to using MetaFrame by comparing the TSE Remote Desktop Protocol (RDP) protocol (the protocol for stand-alone TSE installations) to the Citrix MetaFrame (a TSE add-on) Independent Computing Architecture (ICA) protocol.

Results We conducted three sets of tests to compare ICA to RDP: Sign-on over WAN connections, selected application (Address Book, Item Master, and Requisition Creation) performance over Local Area Network (LAN) connections, and selected application (same as above) performance over WAN connections. ??For WAN sign-on, ICA performance is superior to RDP in most measured categories, and shows a 39% reduction in duration (time to complete task) relative to RDP. ??For OneWorld Applications run over a LAN, ICA slightly outperforms RDP as measured by average duration time. RDP does transmit at a higher rate in most cases, but also transmits much more traffic (bytes), which accounts for the higher RDP duration times. However, based on these application tests, the use of MetaFrame in a LAN environment does not significantly improve performance relative to RDP ??For OneWorld applications run over a WAN, ICA outperformed RDP in all tests when measured in response time alone. In all three tests, RDP generated fewer frames and turns. RDP transferred less data (bytes) in the Item Master test, but more data in the Address Book and Requisition Master tests. RDP and ICA transmission rates are nearly equal for the Address Book test, but ICA shows better transmission rates in the other two tests. Even though ICA performs more efficiently than RDP in these WAN tests, different WAN connection characteristics could lead to varying results. Before purchasing Citrix MetaFrame as an add-on to OneWorld/TSE implementations, you should carefully evaluate your enterprise and assess the need for MetaFrame. Although we expect that most customers should see similar performance to the that reported here, results may vary.

J.D. Edwards Research Series – Baselines

Page 1

Performance Differences between Remote Desktop Protocol (RDP) and Independent Computing Architecture (ICA)

Introduction In September 1998, Worldwide Advanced Technologies Corporate Technology Services team conducted preliminary sizing tests which increased interest in TSE when used in conjunction with OneWorld 1 . Following our initial investigation, a customer indicated interest in further TSE/MetaFrame testing, and participated in the study reported here.

Running TSE Alone and Remote Desktop Protocol (RDP) TSE allows users to run applications on a remote terminal server as if those applications were running on a local desktop. The TSE operating system greatly reduces WAN traffic by only downloading only screen presentations from a terminal server, and only uploading keystrokes and mouse movements from client workstations. TSE uses RDP to communicate between the terminal server and the end user. RDP is specifically designed for use on a LAN and functions adequately in that environment.

Citrix MetaFrame and Independent Computing Architecture (ICA) Based on earlier tests and those reported in this document, we have found that OneWorld customers using TSE can find some benefit to using Citrix MetaFrame, a TSE add-on. Citrix MetaFrame uses a communications protocol called Independent Computing Architecture (ICA) developed specifically for TSE applied over a WAN. Our preliminary tests show that the Citrix MetaFrame ICA protocol is more efficient and functional than the TSE-only RDP protocol. The cost of Citrix MetaFrame has prompted us to evaluate the performance differences between the TSE RDP protocol and Citrix MetaFrame’s ICA protocol. The preliminary tests reported here are intended to give a general idea of the differences before making a decision to purchase this TSE add-on.

J.D. Edwards Research Series – Baselines

Page 2

Performance Differences between Remote Desktop Protocol (RDP) and Independent Computing Architecture (ICA)

Additional Advantages to Using MetaFrame The following are a number of additional benefits to using Citrix MetaFrame. ??MetaFrame provides functionality for all types of Windows and also supports nonWindows clients including DOS, UNIX, Mac OS, Java and OS/2 Warp and a broad range of client hardware including legacy PCs, Pentium PCs, Windows-based terminals, network computers, wireless devices and information appliances. ??MetaFrame dynamically connects users to the network through standard telephone lines, WAN links (T1, T3, 56Kb, X.25), broadband connections (ISDN, Frame Relay, ATM), wireless connections, corporate Intranets, and the Internet. ??MetaFrame offers seamless Windows integration by providing complete access to all local system resources, including local drives, COM ports, full 16-bit stereo audio, cut and paste, drag and drop, and local printers.

Test Environment Our testing environment was developed for a customer who is planning a very complex WAN strategy involving TSE and high-volume data replication. Because our customer wanted near simultaneous tests using the same hardware for both tests, we loaded Citrix MetaFrame on the Terminal Server. However, we believe that MetaFrame had no effect on our RDP results since RDP is an entirely different protocol than ICA and should not be able to communicate with MetaFrame on the Terminal Server. Nevertheless, even when not used by clients, some believe that MetaFrame enhances file management, although we do not have evidence in support of this idea. Enterprise Server Terminal Server Deployment Server Database Client PCs LAN WAN

IBM Netfinity 5500 HP LXR 4 Way Pentium Pro 200, 1 GB RAM HP Vectra 166, 128 MB RAM Oracle 8.04 HP Vectra 166, 128 MB RAM 10baseT 56 kbps

Sign-On Tests The sign-on tests we conducted were designed to measure relative performance for the TSEonly versus TSE/MetaFrame over a 56 kbps WAN connection. These tests measured the time from double-clicking the OneWorld Explorer Desktop icon to the time that the main OneWorld menu appeared. Tests were run manually and are subject to some variability and user error. However, care was taken to be as consistent as possible under these conditions. RDP uses two client connection settings: high-speed for LAN connections, and low-speed for WAN connections. We conducted two tests to evaluate the differences between these two RDP settings relative to ICA. The tests were conducted manually and at different points in time, so the results should be interpreted as relative rather than absolute. One important point to note about the sign-on tests is that, between sessions, ICA protocol caches bitmaps on the client’s hard disk. For sign-on this is significant because, the OneWorld “globe” logo (that appears prior to entering the program) is cached after a user’s initial sign-on.

1

Running OneWorld under Windows Terminal Server Edition, September, 1998

J.D. Edwards Research Series – Baselines

Page 3

Performance Differences between Remote Desktop Protocol (RDP) and Independent Computing Architecture (ICA)

This can account for some of the performance differences you see between TSE-only and TSE/MetaFrame.

J.D. Edwards Research Series – Baselines

Page 4

Performance Differences between Remote Desktop Protocol (RDP) and Independent Computing Architecture (ICA)

Sign-On Test One – RDP High Speed LAN Connection We first tested sign-on performance for the two protocols over a WAN connection. These tests measured the time from double-clicking the OneWorld Explorer Desktop icon to the time that the main OneWorld menu appeared. RDP uses two client connection settings: high-speed for LAN connections, and low-speed for WAN connections. For test one, we used the RDP client LAN high-speed connection. The differences between the two protocols are shown below. ICA outperformed RDP in all the important categories. Most significantly, the ICA client took 40 percent less time and generated 80 percent less byte traffic than the RDP client did.

ICA vs. RDP Sign-On Comparison Tests Task

(RDP client set to high speed LAN connection) RDP Sign-On ICA Sign-On Difference % Difference

Task Duration Total Frames Total Bytes Avg. Frame Size Avg. Frames/sec Payload Bytes Payload % Total Overhead Bytes Overhead % Of Total Avg. bps Avg. Bps Total App. Turns TCP Turns Frames/App. Turn Bytes/App. Turn Client Processing Time Client Frames Sent Client Avg. Frame Size Sent Client Bytes Sent Client Bits Sent Client Payload Bytes Sent Client Payload % Sent Client Overhead Bytes Sent Client Overhead % Server Processing Time Server Frames Sent Server Avg. Frame Size Server Bytes Sent Server Bits Sent Server Payload Bytes Sent Server Payload % Sent Server Overhead Bytes Sent Server Overhead % Sent

J.D. Edwards Research Series – Baselines

29.36 288 102,722 356.67 9.81 86,438 84.15% 16,284 15.85% 27,987.74 3,498.47 20 66 14.4 5,136.10 16.97 138 94.45 13,034 104,272 5,069 38.89% 7,965.00 61.11% 9.8 150 597.92 89,688 717,504 81,369 90.72% 8,319.00 9.28%

17.85 182 20,755 114.04 10.2 10,483 50.51% 10,272 49.49% 9,304.05 1,163.01 14 43 13 1,482.50 7.75 99 62.23 6,161 49,288 626 10.16% 5,535.00 89.84% 6.98 83 175.83 14,594 116,752 9,857 67.54% 4,737.00 32.46%

-11.51 -106 -81967 -242.63 0.39 -75955 -0.3364 -6012 0.3364 -18683.7 -2335.46 -6 -23 -1.4 -3653.6 -9.22 -39 -32.22 -6873 -54984 -4443 -0.2873 -2430 0.2873 -2.82 -67 -422.09 -75094 -600752 -71512 -0.2318 -3582 23.18%

-39.20% -36.81% -79.79% -68.03% 3.98% -87.87% -39.98% -36.92% 212.24% -66.76% -66.76% -30.00% -34.85% -9.72% -71.14% -54.33% -28.26% -34.11% -52.73% -52.73% -87.65% -73.88% -30.51% 47.01% -28.78% -44.67% -70.59% -83.73% -83.73% -87.89% -25.55% -43.06% 249.78%

Page 5

Performance Differences between Remote Desktop Protocol (RDP) and Independent Computing Architecture (ICA)

Sign-On Test Two – RDP Low Speed WAN Connection For Test Two, we changed the RDP client connection setting to “low-speed” to evaluate RDP performance for a WAN environment. In all other ways, this is the same as Sign-on Test One. These results show a slight improvement in RDP relative to ICA in many categories, but overall, ICA still outperforms RDP.

ICA vs. RDP Sign-On Comparison Tests Task

(RDP client set to low speed LAN connection) RDP Sign-On ICA Sign-On Difference % Difference

Task Duration Total Frames Total Bytes Avg. Frame Size Avg. Frames/sec Payload Bytes Payload % Total Overhead Bytes Overhead % Of Total Avg. bps Avg. Bps Total App. Turns TCP Turns Frames/App. Turn Bytes/App. Turn Client Processing Time Client Frames Sent Client Avg. Frame Size Sent Client Bytes Sent Client Bits Sent Client Payload Bytes Sent Client Payload % Sent Client Overhead Bytes Sent Client Overhead % Server Processing Time Server Frames Sent Server Avg. Frame Size Server Bytes Sent Server Bits Sent Server Payload Bytes Sent Server Payload % Sent Server Overhead Bytes Sent Server Overhead % Sent

21.75 218 66,384 304.51 10.02 54,138 81.55% 12,246 18.45% 24,415.62 3,051.95 15 45 14.53 4,425.60 12.19 98 85.1 8,340 66,720 2,676 32.09% 5,664.00 67.91% 9.31 120 483.7 58,044 464,352 51,462 88.66% 6,582.00 11.34%

13.24 110 16,062 146.02 8.31 9,900 61.64% 6,162 38.36% 9,705.81 1,213.23 13 13 8.46 1,235.54 6.74 50 60.26 3,013 24,104 175 5.81% 2,838.00 94.19% 6.21 60 217.48 13,049 104,392 9,725 74.53% 3,324.00 25.47%

-8.51 -108 -50,322 -158.5 -1.71 -44,238 -19.91% -6,084 19.91% -14,709.81 -1,838.73 -2 -32 -6.07 -3,190.06 -5.45 -48 -24.84 -5,327 -42,616 -2,501 -26.28% -2,826.00 26.28% -3.1 -60 -266.22 -44,995 -359,960 -41,737 -14.13% -3,258.00 14.13%

-39.13% -49.54% -75.80% -52.05% -17.10% -81.71% -24.41% -49.68% 107.91% -60.25% -60.25% -13.33% -71.11% -41.78% -72.08% -44.74% -48.98% -29.19% -63.87% -63.87% -93.46% -81.89% -49.89% 38.70% -33.28% -50.00% -55.04% -77.52% -77.52% -81.10% -15.94% -49.50% 124.60%

Summary of Sign-On Tests These preliminary sign-on tests indicate ICA superiority over RDP in most measured categories. Compared to the high-speed RDP setting, the RDP low-speed setting seems to improve performance slightly relative to ICA. However, for both the low- and high-speed RDP settings, sign-on duration times and network traffic are more favorable for ICA.

J.D. Edwards Research Series – Baselines

Page 6

Performance Differences between Remote Desktop Protocol (RDP) and Independent Computing Architecture (ICA)

Scripted Application Tests – LAN For our LAN (10baseT) applications tests, we ran scripts to simulate the application process and then captured Application Expert data. In the scripted application tests, the first run was conducted before bitmap caching, while the second and third occurred after caching. Results from each captured test are shown along with average results for each test.

Address Book LAN Tests These tests are based on a simple Address Book entry. Average ICA response time was lower than for RDP, largely a result of the lower number of bytes transferred by ICA. However, RDP showed higher transmission rates than ICA, but the rate was not enough to overcome the higher number of bytes transferred. Consequently, RDP showed a somewhat higher response time.

ADDRESS BOOK LAN TESTS ICA Capture 1 Capture 2 Duration (sec.) 54.03 42.61 Frames 1,203 1,185 App. Frames 847 832 %App. Frames 70% 70% Bytes 125,119 111,769 Avg. Frame 104 94 App. Turns 173 165 TCP Turns 204 190 Frm./App. Turn 6.95 7.18 Bytes/App. Turn 723 677 Rate (bits/sec.) 18,527 20,984 Date Captured 11/12/98 11/12/98 15:59 16:01

Capture 3 42.11 1,299 918 71% 117,711 91 178 218 7.30 661 22,362 11/12/98 16:02

J.D. Edwards Research Series – Baselines

RDP Average Capture 1 Capture 2 Capture 3 Average 46.25 59.69 46.24 42.11 49.35 1,229 942 807 797 849 866 662 580 573 605 70% 70% 72% 72% 71% 118,200 236,528 180,151 179,027 198,569 96 251 223 225 233 172 131 120 115 122 204 192 169 162 174 7.14 7.19 6.73 6.93 6.95 687 1,806 1,501 1,557 1,621 20,624 31,703 31,170 34,011 32,295 11/12/98 11/12/98 11/12/98 15:28 15:30 15:31

Page 7

Performance Differences between Remote Desktop Protocol (RDP) and Independent Computing Architecture (ICA)

Item Master LAN Tests For the Item Master tests, average ICA response time was lower than for RDP, largely a result of the lower number of bytes transferred by ICA. Even though RDP showed higher transmission rates than ICA, the rate was not enough to overcome the higher number of bytes transferred. Consequently, RDP showed a somewhat higher response time.

Item Master LAN Tests ICA

RDP

Capture 1 Capture 2 Capture 3 Average Capture 1 Capture 2 Capture 3 Average Duration (sec.) 57.68 51.97 50.32 53.32 58.86 52.79 56.70 56.12 Frames 2,148 2,192 1,988 2,109 1,187 1,086 1,128 1,134 App. Frames 1517 1550 1418 1495 847 773 807 809 %App. Frames 71% 71% 71% 71% 71% 71% 72% 71% Bytes 216,171 217,164 204,821 212,719 292,681 281,370 286,397 286,816 Avg. Frame 101 99 103 101 247 259 254 253 App. Turns 246 234 255 245 159 136 155 150 TCP Turns 437 479 386 434 246 227 228 234 Frm./App. Turn 8.73 9.37 7.80 8.63 7.47 7.99 7.28 7.58 Bytes/App. Turn 879 928 803 870 1,841 2,069 1,848 1,919 Rate (bits/sec.) 29,981 33,426 32,561 31,989 39,777 42,643 40,408 40,942 Date Captured 11/12/98 11/12/98 11/12/98 11/12/98 11/12/98 11/12/98 16:04 16:06 16:07 15:36 15:37 15:39

J.D. Edwards Research Series – Baselines

Page 8

Performance Differences between Remote Desktop Protocol (RDP) and Independent Computing Architecture (ICA)

Requisition Creation LAN Test For the Requisition Creation tests, the response time for both protocols was very similar. Here ICA showed a lower number of bytes transferred and a lower transmission rate. Because RDP is able to achieve higher transmission rates, the two average duration times are nearly equal even though RDP transmits more bytes.

Requisition Creation LAN Tests ICA

RDP

Capture 1 Capture 2 Capture 3 Average Capture 1 Capture 2 Capture 3 Average Duration (sec.) 40.95 33.66 34.98 36.53 34.19 38.74 34.07 35.67 Frames 1,349 1,234 1,170 1,251 559 655 578 597 App. Frames 944 862 822 876 399 463 410 424 %App. Frames 70% 70% 70% 70% 71% 71% 71% 71% Bytes 120,100 104,998 99,958 108,352 131,958 156,507 127,066 138,510 Avg. Frame 89 85 85 87 236 239 220 232 App. Turns 116 121 118 118 84 89 89 87 TCP Turns 332 281 267 293 103 132 105 113 Frm./App. Turn 11.63 10.20 9.92 10.58 6.65 7.36 6.49 6.83 Bytes/App. Turn 1,035 868 847 917 1,571 1,759 1,428 1,586 Rate (bits/sec.) 23,463 24,955 22,860 23,760 30,877 32,323 29,837 31,012 Date Captured 11/12/98 11/12/98 11/12/98 11/12/98 11/12/98 11/12/98 16:09 16:13 16:14 15:42 15:43 15:44

Summary of LAN Test Data From these LAN application tests, a couple facts are evident. ??ICA slightly outperforms RDP in these LAN application tests, as measured by average response time (duration). ??RDP transmits more network traffic (bytes) but it makes up some of the difference by transferring data at a higher rate than ICA. Based on these application tests, the use of MetaFrame in a LAN environment does not significantly improve performance relative to RDP.

J.D. Edwards Research Series – Baselines

Page 9

Performance Differences between Remote Desktop Protocol (RDP) and Independent Computing Architecture (ICA)

Scripted Application Tests – WAN For our WAN (56 kbps) applications tests, we ran scripts to simulate the application process and then captured Application Expert data. In the scripted application tests, the first run was conducted before bitmap caching, while the second and third occurred after caching. Results from each captured test are shown along with average results for each test.

Address Book WAN Test In our Address Book tests, we found that RDP generated fewer frames and turns than ICA, but transmitted more data (bytes). The transmission rates were about equal for RDP and ICA. The reduction in transmitted data for ICA resulted in a lower average ICA duration (time to complete task) relative to RDP duration.

Address Book WAN Test ICA

RDP

Capture 1 Capture 2 Capture 3 Average Capture 1 Capture 2 Capture 3 Average Duration (sec.) 58.01 54.68 55.45 56.05 60.61 68.06 60.92 63.20 Frames 1,386 1,379 1,288 1,351 920 882 855 886 App. Frames 974 962 911 949 643 610 592 615 %App. Frames 70% 70% 71% 70% 70% 69% 69% 69% Bytes 130,527 123,746 118,346 124,206 133,305 160,184 127,516 140,335 Avg. Frame 94 90 92 92 145 182 149 159 App. Turns 174 168 181 174 149 122 133 135 TCP Turns 237 245 207 230 119 128 124 124 Frm./App. Turn 7.97 8.21 7.12 7.77 6.17 7.23 6.43 6.61 Bytes/App. Turn 750 737 654 714 895 1,313 959 1,055 Rate (bits/sec.) 18,000 18,105 17,074 17,726 17,596 18,829 16,746 17,724 Date Captured 11/11/98 11/11/98 11/11/98 11/11/98 11/11/98 11/11/98 16:01 16:04 16:06 15:48 15:54 15:56

Item Master WAN Test In the Item Master WAN tests, RDP showed lower frames, turns, and transmitted bytes. However, here the transmission rate for ICA was far superior to RDP. Consequently, ICA showed a lower duration time than RDP.

Item Master WAN Test ICA

RDP

Capture 1 Capture 2 Capture 3 Average Capture 1 Capture 2 Capture 3 Average Duration (sec.) 84.47 72.06 70.1 75.54 94.8 77.64 81.21 84.55 Frames 2,931 2,501 2,706 2,713 1,262 1,173 1,138 1,191 App. Frames 2044 1766 1924 1911 886 832 795 838 %App. Frames 70% 71% 71% 70% 70% 71% 70% 70% Bytes 266,607 237,415 247,356 250,459 239,771 194,946 188,543 207,753 Avg. Frame 91 95 91 92 190 166 166 174 App. Turns 270 269 289 276 187 178 163 176 TCP Turns 657 531 572 587 166 170 160 165 Frm./App. Turn 10.86 9.30 9.36 9.84 6.75 6.59 6.98 6.77 Bytes/App. Turn 987 883 856 909 1,282 1,095 1,157 1,178 Rate (bits/sec.) 25,249 26,356 28,229 26,611 20,235 20,087 18,574 19,632 Date Captured 11/11/98 11/11/98 11/11/98 11/11/98 11/11/98 11/11/98 16:24 16:26 16:27 16:31 16:32 16:35

J.D. Edwards Research Series – Baselines

Page 10

Performance Differences between Remote Desktop Protocol (RDP) and Independent Computing Architecture (ICA)

Requisition Creation WAN Test In the WAN Requisition Creation test, RDP again shows lower frame and turn numbers. In this instance, ICA transmits slightly fewer bytes and does so at a higher transmission rate than RDP. This test shows the largest difference in duration times, with ICA demonstrating much better response.

Requisition Creation WAN Test ICA Capture 1 Capture 2 Duration (sec.) 48 41.76 Frames 1,276 1,279 App. Frames 891 895 %App. Frames 69.83% 69.98% Bytes 110,845 112,800 Avg. Frame 86.87 88.19 App. Turns 126 127 TCP Turns 284 279 Frm./App. Turn 10.13 10.07 Bytes/App. Turn 879.72 888.19 Rate (bits/sec.) 18,474.53 21,609.14 Date Captured 11/11/98 11/11/98 16:59 17:00

Capture 3 44.44 1,286 893 69.44% 112,059 87.14 128 275 10.05 875.46 20,170.84 11/11/98 17:01

RDP Average Capture 1 Capture 2 Capture 3 Average 44.73 70.2 55.92 47.86 57.99 1,280 745 699 657 700 893 510 485 454 483 70% 68.46% 69.38% 69.10% 69% 111,901 143,146 105,795 100,201 116,381 87 192.14 151.35 152.51 165 127 104 104 98 102 279 110 105 110 108 10.08 7.16 6.72 6.7 6.86 881 1,376.40 1,017.26 1,022.46 1,139 20,085 16,312.70 15,135.03 16,749.42 16,066 11/11/98 11/11/98 11/11/98 16:49 16:51 16:52

Summary of WAN Test Data The WAN results are summarized here. ??In all three tests, RDP generated fewer frames and turns, factors that would be significant on high-latency connections. ??RDP transferred less data (bytes) in the Item Master test, but more data in the Address Book and Requisition Master tests. ??Although transmission rates are nearly equal for the Address Book test, ICA shows better transmission rates in the other two tests. ??The bottom line is that, based on duration time alone, ICA outperformed RDP in all tests. The WAN application tests reported here indicate that MetaFrame performs more efficiently than RDP in a WAN environment. However, with different WAN connection characteristics, these results could vary. Especially with high latency connections you might see closer response times between the two protocols.

J.D. Edwards Research Series – Baselines

Page 11

Performance Differences between Remote Desktop Protocol (RDP) and Independent Computing Architecture (ICA)

Conclusions and Recommendations The factors outlined in this report indicate some performance advantages to using Citrix MetaFrame when running OneWorld under TSE. There is additional MetaFrame functionality mentioned in this report that also increases its value to the customer. These additional functions include support for all Windows and nonWindows clients, support for most network connections and protocols, cost-effective management and sizing for growing enterprises, and seamless Windows integration and functionality for clients. For LAN-based OneWorld customers not wanting to run ordinary fat clients, Citrix MetaFrame ICA protocol shows slightly better performance than TSE-alone RDP protocol. This slight performance difference between RDP and ICA is not enough to warrant the addition of MetaFrame for LAN users. However, for customers using MetaFrame for WAN users when both LAN and WAN users are on the system, it may be advantageous to employ MetaFrame for all users (both LAN and WAN) to avoid any “mix and match” confusion. Finally, before purchasing Citrix MetaFrame as an add-on to OneWorld/TSE implementations, you should carefully evaluate your enterprise and assess the need for MetaFrame. Although we expect that most customers should see similar performance to the results reported here, results may vary.

Contact Bernard Douthit J.D. Edwards Worldwide Advanced Technologies (303) 334-1096 [email protected]

J.D. Edwards Research Series – Baselines

Page 12

SO F T W A R E

FOR A

CHANGING

W O R L D?

J.D. Edwards World Solutions Company World Headquarters One Technology Way Denver, Colorado 80237, USA 800 727 5333 / 303 334 4000

J.D. Edwards World Solutions Company Serving Latin America and the Caribbean 800 Douglas Entrance Coral Gables, Florida 33134, USA 305 442 7800

J.D. Edwards U.K. Ltd. Serving Europe, the Middle East, and Africa Oxford Road Stokenchurch, High Wycombe Buckinghamshire HP14 3AD, United Kingdom 44 1494 682700

J.D. Edwards Hong Kong Limited Serving Asia Pacific Unit A, 24 th Floor, Entertainment Building 30 Queen’s Road Central, Hong Kong 852 2844 7400

www.jdedwards.com

The materials contained herein are summary in nature, subject to change, and intended for general information only. J.D. Edwards is a registered trademark of J.D. Edwards & Company. The names of all other products and services of J.D. Edwards used herein are trademarks or registered trademarks of J.D. Edwards World Source Company. All other product names used are trademarks or registered trademarks of their respective owners. ? J.D. Edwards World Source Company 1998 One or more U.S. Patent Applications are pe nding for inventions used in the production of OneWorld.