JUDELOGO
Product List
System Design Tool - JUDE/Professional
UML Modeling Tool - JUDE/Community
Spec Table
System Requirements
Compatibility
Mind Map
JUDE API
XMI Input/Output
JUDE/Server
Change Vision, Inc.
Note on JUDE/Server

Number of people to use JUDE/Server

JUDE/Server is recommended for use in a small projects that have been handled by less than 10 people. JUDE/Server is not recommended for a big project team because there would be waiting time for each user to access the server in a big team. With JUDE, users are updating project models in local machines, and are not constantly connected to the server. Accessing the server is only required when checking in/out. The server can only operate by one user at the time, so the rest of the users would have to wait until one user finishes its process completely.

Having many accesses at the same time could cause performance problems. Also, if you are handling a large size of model and you are using not powerful machines, they can affect the server performance, too.

JUDE/Server System Requirements

Please refer to Requirements.
Note : Less of Memory Space of Server machine could cause performance problems.

Recommended file size to use on JUDE/Server

It is recommended to use Project File that is less than 1MB on JUDE/Server.

*Note1 Information about the size of Project file
  Please refer to the following table regarding project scale and file sizes.

Project Size Project Contents
Default Project 76KB 3 Packages, 2 Empty Diagrams, 58 Classes
snapshot.jude Project 34KB 6 Packages, 6 Diagrams, 23 Classes
1MB@Project 1MB About 500 Diagrams

*NOTE2 About size of Project file
  These sizes are data that have saved in default setting of JUDE.
  Data are compressed when saving by JUDE default setting.

Note on Use

  • When more than one client try to access to server at the same time, it brings waiting time for users except one who is currently in process. For instance, there are 2 clients trying to check out at the same time, one person has to wait until another one's procedure completes successfully. The server runs in proper condition even if more than one client have been logged on or modifying the data at the same time after checking out. However when one more client try to check-in / check-out / log-in / log-out at the same time, the server condition goes busy and the rest of the clients except one in progress have to wait until one completes the procedure.
  • It is not recommended to use bigger size of project file than 1MB. JUDE/Server memory is used for whole models of all currently logged on users. By doing so, the server consumes some memories and could cause performance problems. In case the project team handles the big scale model, that means clients will have to open the project on clients machine, so you have to be aware that client's machine also should be high performance as much as server machine is.
  • It is possible to check out only a part of project. However, if there are packages that have interaction to the selected part, you have to check out them together.
  • Collaboration with JUDE/Server is out of our support.
    It is recommended to take backup regularly to prevent form losing your important data.

Tips to reduce the waiting time

  • Try not to have more than one client check in/out at the same time.
  • Make the project size smaller.
  • Use the high performance machine for the server. Make more plenty of memory space.
  • Check off [Project History] function. (This will cut off the export of a log file to the server)
  • Try not to log in more than one project in one server at the same time.

Tips to use large size models on JUDE/Server

  • Use high performance machines for both client and server.(More than CPU2.5GHz and Memory 2.0G)
  • Make VM Memory setting bigger on client machine. (Default setting : -Xms16m -Xmx384m -Xss2m)
  • Make Tomcat VM Memory bigger than what Server Guide indicates. (Initial memory pool:16, Maximum memory pool:512, Thread stack size:2048)
  • Make Xss or Thread stack size bigger, when it appears "StackOverFlow".
  • When Checking out, try to check out a part of package not whole package.


Performance on JUDE/Server

Here is the result of a performance survey with JUDE/Server.

Machine Specifications

¡Server
CPU 2.8GHz
Memory 2.0GB
Java J2SDK 1.4.2_10
Tomcat Tomcat 5.0.28
  Initial memory pool: 16MB
  Maximum memory pool: 512MB
  Thread stack size: 2,048KB

JUDE/Server 2.4.1

¡Clients
CPU 2.8GHz
Memory 2.0GB
Java J2SDK 1.4.2_10
  -Xms16m -Xmx384m -Xss2m

JUDE/Enterprise 2.4.3
JUDE/Server 3.x and JUDE/Professional 3.x. have same performance as much as above versions.

Model

¦ Each model includes all kinds of diagrams and model elements.
Model Size of Model Diagram
Model A 113KB
21
Model B 303KB
101
Model C 500KB
158
Model D 803KB
more than 200
Model E 1,616KB

Model F 3,209KB

Performance Result on Clients

Model File
Open
File
Save
Alias
Switch
Model A 2 sec
1 sec
-
Model B 2 sec
3 sec
5 sec
Model C 3 sec
6 sec
9 sec
Model D 7 sec
7 sec
15 sec
Model E 17 sec
16 sec
60 sec
Model F 60 sec
30 sec
196 sec


Model Result
Model A Works properly
Model B Works properly
Model C It seems to be working slower during following performances.
  • Switching alias.
  • Cloning packages that contain a lot of models.
  • Deleting packages that contain a lot of models.
Model D It apparently works slower during following performances.
  • Opening files.
  • Closing files.
Model E It is losing JUDE's lightweight touch. It is clearly slow during following performances.
  • Opening files.
  • Closing files.
It is very slow especially during [Switching Alias]. It seems like that this is a limit size we can recommend.
Model F It is very difficult to operate. It is not recommended to use this size of file but it still can be used. You just need a patience to use it.
¡About a size of one project
[Switching Alias] takes the longest to operate compared to other operations when handling big size of models. In spite of using the high performance machine (CPU:2.8GHz, Memory:2.0G) for server, when client opens *.jude file that is bigger than 5M, the response is very slow.

Information about Collaboration running time

This is surveyed in a same condition as above.

[Condition]
1. Same machine is used as Server and Client for this survey.
2. [Project History] on JUDE/Server is set to default setting.
3. Run the server in initial condition.



Model A
3 Log in to Server
4 Add Model A onto Server
18 Sec
5 Log out from Server

6 Log in to Model A

7 Check out whole model of Model A
8 Sec
8 Edit one topic on Mind Map of Model A, Then Check In.
5 Sec
9 Move all elements of whole diagrams in Model A then Check in.
6 Sec
10 Log out from Model A



Model B
11 Log in to server

12 Add Model B onto Server
23 Sec
13 Log out from Server

14 Log in Model B

15 Check out Model B
25 Sec
16 Edit one topic on Mind Map of Model B, Then Check In.
20 Sec
17 Move all elements of all diagrams in Model B, then Check in
22 Sec
18 Log out Model B



Model C
19 Log in to Server

20 Add Model C onto Server
33 Sec
21 Log out from Server

22 Log in to Model C

23 Check out Model C
33 Sec
24 Edit one topic on Mind Map of Model C, Then Check In.
33 Sec
25 Move all elements of whole diagrams in Model C then Check in.
36 Sec
26 Log out from Model C



Model D
27 Login to Server

28 Add Model D onto server
54 Sec
29 Log out from Server

30 Log in to Model D

31 Check out from Model D
60 Sec
32 Edit one topic on Mind Map of Model D, Then Check In.
54 Sec
33 Log out from Model D



Model E
34 Log in to Server

35 Add Model E onto server
127 Sec
36 Log out from Server

37 Log in to Model E

38 Check out Model E
152 Sec
39 Edit one topic on Mind Map of Model E, Then Check In.
135 Sec
40 Log out from Model E



Model F
41 Log in Server

42 Add Model F on the Server
184 Sec
43 Log out from Server

44 Log in Model F

45 Check out Model F
368 Sec
46 Edit one topic on Mind Map of Model F, Then Check In.   @¨   @Error : out of memory
  ¦This might be solved by making size of Maximum memory pool of TOMCAT bigger.
47 Log out from Model F



About us    Privacy Statement   Copyright(C) 2006-2008 Change Vision, Inc. All rights reserved.