PLATEFORME - Installation & Deployment (EN) > HOPEX Application Server (HAS) Architecture Overview > Sizing Physical Infrastructure
6. Sizing Physical Infrastructure
 
Disclaimer
Hardware sizing
Server configuration
For Production
Other server environments
How to calculate maximum simultaneous users
Multiple instances
Public vs Private Workspace
Making the right choice
6.1. Disclaimer
The following sizing is based on our regular benchmark and load testing performed by the R&D. It is made based on the following assumptions:
• Smallest sizing: Possible for few concurrent users
• Small deployment: Up to 15 concurrent users
• Medium deployment: Up to 40 concurrent users
• Large deployment: Count one HAS Server for each 40 concurrent users' group.
 
This infrastructure can be:
• Physical server
• Virtual server: In this context the physical underlying infrastructure must be sized enough to support all running virtual servers.
 
We recommend a dedicated server for the HAS Server layer.
MEGA has made reasonable efforts to ensure the quality, accuracy, and validity of the performance benchmarking resulting in this sizing. Changes in any of the server's parameters might cause a positive or negative effect on the user experience and performances.
 
6.2. Hardware sizing
6.2.1. Server configuration
This sizing is based on the following hardware configurations. All HDD are of SSD type in these configurations.
 
Sizing
CPU Core
RAM
HDD
S1
2
8
100 Gb
S2
4
16
128 Gb
S3
8
32
128 Gb
S4
16
64
128 Gb
6.2.2. For Production
• Users mentioned in this table are maximum number of simultaneously connected users. (see below for calculation rule)
• In the cell the “S+number” represents the server configuration to choose.
• In bold are the preferred choices
 
 
 
Configuration
 
Max simultaneous users:
<2
<7
<16
<41
>40
1
Single Laptop
HAS Application Server
SQL Server
S2
 
 
 
 
2
Single Server
Smallest sizing
IIS Web Server
HAS Application Server
SQL Server
 
S2
S3
 
 
3.1
Two servers
Medium Deployment
IIS Web Server
HAS Application Server
 
S2
S2
S3
 
SQL Server
 
S1
S2
S2
 
3.2
Two servers
Medium Deployment
IIS Web Server
 
S1
S1
S1
 
HAS Application Server
SQL Server
 
S2
S3
S4
 
4
Three servers
Medium Deployment
IIS Web Server
 
 
S1
S1
 
HAS Application Server
 
 
S2
S3
 
SQL Server
 
 
S2
S3
 
5
Cluster/Farms
Large Deployment
IIS Web Servers
 
 
S1
S1
S1
HAS Application Servers*
 
 
S2
S3
S3
SQL Server
 
 
S2
S3
S3
* Add one server for each additional group of users 40 users.
 
6.2.3. Other server environments
• For development: use Single Server with Sizing 2
• For training: 10 concurrent users, use Single Server with Sizing 3
• For pre-production: same infrastructure pattern as production with Sizing 2
6.2.4. How to calculate maximum simultaneous users
The maximum number of simultaneous users depends on the type of users:
• Main users: these are users using the tool on a regular basis. They have tasks to perform that can take several hours.
• Contributors/Viewers users: these are users that consume information and have limited production contribution. Their usage is punctual over the weeks with limited time spent when they connect.
Complete the following table to find your number of maximum simultaneous users.
 
License users
Number
Formula
Total
Mains users
 
RoundUp (Nb / 4 )
 
Contributors/Viewers
 
RoundUp (Nb / 100)
 
Maximum simultaneous users:
 
 
Example:
You have 5 process modelers, 10 portfolio managers, 40 application owners, 100 viewers. I will then have:
 
License users
Number
Formula
Total
Mains users
15
RoundUp (Nb / 4 )
4
Contributors/Viewers
140
RoundUp (Nb / 100)
2
Maximum simultaneous users:
6
 
You can choose a single server or two server deployment type. In that context the preferred deployment type is the one highlighted in bold.
 
6.2.5. Multiple instances
The sizing proposed here is done for only 1 HAS Instance on the server. Should you be in a multi-instance scenario you need to adjust RAM consumption accordingly.
 
Count minimum 5 Go additional RAM for each new Instance. The needed RAM also depend on maximum concurrent users.
6.2.6. Public vs Private Workspace
In most of the desktops, HOPEX users work in public workspaces, i.e. their actions are automatically saved (within 5 min).
 
 
Multi-Session (MS)
Single Session (SS)
 
Public Workspace
 
Default – recommended
Not supported /
Not Available
 
Private Workspace
 
Not supported /
Not Available
V3/V4: behavior
V5: possible
 
Changing the behavior from public workspace to private workspace has a direct impact on sizing.
You must adjust RAM consumption: count 1Go of RAM for each additional concurrent user.
Example: You change 10 BPA Modeler into private workspace (SS)
With the new behavior you need to add 10Go of RAM to the server.
6.2.7. Making the right choice
Refer to the decision tree to choose the deployment type.
Select the preferred configuration sizing among the deployment type.