Dear Martijn,

Please find inputs to your questions :

• Are those modules stable enough to use them as main blocks in the computing logic?

The modules you choose are the most used one and even if not yet completely mature, the increasing number of users will help to identify and solve any instability that could be met.

• Are there any well known issues with any of them?

As a potential issue I know at least that the hadoop version used in the current version of Orion is a bit old and this you may be limited in using advanced features available in newer versions. In general, check knowledge base in wiki<https://forge.fiware.org/plugins/mediawiki/wiki/fiware/index.php/FI-WARE_Architecture>, in StackOverflow.com<http://stackoverflow.com/search?q=fiware> as well as the e-learning portal<http://edu.fi-ware.org/>.

• Are there any possible incompatibility between those Fiware modules?

You plan to use them following the most natural architecture in FIWARE, as you can check in the wiki. Getting the enablers working together in an integrated way is exactly the purpose of FIWARE

• Are there any architectural mistakes in our design?

No. But implementing the whole may be ambitious... Think about incremental steps

• Is there any company that is actively using those modules for context awareness and complex event processing? If yes, for which kind of activities?

I do not have any tangible proof but heard that Telefonica is using its enabler in the context of platform for smart cities. Several other cities are now announcing that they will deploy similarly (such as Amsterdam). However, these are not yet in production as far as I know (but I may be wrong). Anyway, you can check that many enablers are in fact based on strong 3rd party open-source software (such as hadoop) which in some way provide additional insurance on their stability.

• Talking about actual resources, do Orion Pub/Sub Broker and Proteus require a lot of memory and computing resources? Is it feasible to let them work in two tiers but with single instances (without horizontal clustering)? Is there any benchmark on resources needed and performances?

I do not know any benchmark on resources needed and performance and will investigate this. The fiware lab trial account that you may have created would likely not be sufficient for intensive test. However, new 'community' account will be available from Feb. 24th with increased level of resources to satisfy usual needs.

• About the security management suite: how confident could we be on the maturity of the implementation? Who grants for it?

The security monitoring part is done by Thales which is really committed in the project and a serious company when it comes to security. In addition, security enablers basically use the well-known Oauth2 protocol as baseline. One issue could be the version of 3rd party software used as foundation for some of the enablers: these version may not be the latest one and may thus miss some security patches.

• Do you suggest us to trust these modules from a security point of view?

I cannot go further the above statement. I will investigate if any security audit is made.

Also, you may want to check the security enablers made by the fi-star project (www.-fi-star.eu<http://www.-fi-star.eu>) in the context of health applications.

Kind regards
Franck

_______________________________________________
Fiware-fiche-coaching mailing list
Fiware-fiche-coaching@lists.fi-ware.org
https://lists.fi-ware.org/listinfo/fiware-fiche-coaching