The easiest way to define your own modeling tools!
Obeo Designer Team Server
Disclaimer: The administration server for Obeo Designer Team Server is currently installed in Team for Capella Server.
Obeo Designer Team Server deployment recommendations
Operating System
The Obeo Designer Team server should be deployed on a Linux x86_64 system with a JRE x86_64 version 8 or later.
Network
Latency: Client and Team Server
It is recommended to provide a network with the lowest possible latency between the client and the server: in the order of 1 to 10 ms for a round-trip.
Latency: Team server and DB server
It is strongly recommended that the Team server and the DB server are located on the same physical server as latency between the Team server and DB server will impact greatly the overall performances of the solution. As such the best performing deployment is achieved by using the H2 database in embedded mode.
If there is a requirement on the database that prevents from using H2, make sure that the latency is as low as possible.
Network stability
VPN are not recommended (it is a latency factor) as well as other network elements that could drop connections which are more or less inactive. As such wireless connection are also not recommended as any loss of connectivity might lead to instability in the product and loss of data. However, if a network element of this kind is mandatory, an SSH tunnel could be used as a workaround to avoid client/server disconnections.
Server isolation
It is strongly discouraged to deploy the server on a public WAN. Obeo Designer should be the only way to edit the information stored in the database.
Scalability and size of models
Scability and performances are highly dependent on the design of the domain metamodel, the implementation of this metamodel and the Viewpoint Specification Models. The following figures are given with an Ecore model and the EcoreTools tooling which applies the Sirius best practices.
The minimum physical memory dedicated to the Team server is 4 GB for a deployment where the expected model size is in the order of 300 000 model elements. The heap memory available for the server should be increased to support bigger models 8GB should support 600 000 model elements.
The memory usage of the clients will increase when the model which is shared among the clients grows as such these resources might need to be increased for larger models with 8GB being expected for models with 600 000 elements (the exact value might vary depending on the amount of information each model element holds).
The latency of end user operations requiring a full analysis of the model increase as the model grow, this includes : opening and closing a project, deleting model elements and representations, launching a transformation or a code generation. Opening a project (and hence collecting the model from the network) might take up to 1 min for a model with 500K elements.
Models having 1 000 000 model elements are the considered the upper limit for a Collaborative Server deployement.
A given server is expected to be used by 10 to 20 users simultaneously depending on their level of activity.
Disclaimer
Notwithstanding what was stated previously, Obeo Designer and Obeo Designer Team products are not warranted to run without any error or interruption. Obeo does not make any warranty regarding the statements that are under the chapter «Obeo Designer Team Server deployment recommendations», this chapter is provided for information purposes.
You aknowledges and accepts the risks involved in using these products which could include without limitation, down time, loss of connectivity or data, system crashes, bad performances or performance degradation.