Present: Tony Cass (IT/CS), Frederic Chapron (IT?CS, Notes), Francois Duval (EN/EL), Pedro Martel (GS/ASE), Derek Mathieson (GS/AIS), Rui Nunes (GS/ASE), Peter Sollander (BE/OP)<br>
<br>
Tony made an introduction on the objective of the meeting&nbsp;:<br>
<ul>
    <li>
        Approval a key requests which looks stupid for Tony as Peter and Fran&ccedil;ois knows better who are in their team</li>
    <li>
        Today there is a large distribution of keys which are not under control anymore.</li>
</ul>
So, have a solution based on electronic keys looks interesting.<br>
&nbsp;<br>
Rui presentation:<br>
Role based authentication and key locks are two different subjects. Rui&rsquo;s presentation is only on the first topic.<br>
Presentation of the various options.<br>
SALTO = there are 200 of them at CERN, The cost is on the lock but not on the card.<br>
LOCKEN = the cost is on the key not on the lock. The lock does not need any power, it is a passive device. The key hosts an electronic device with a battery that allows to open/close the lock. The logs are downloaded to the server when the key is connected to the battery loader.<br>
EVOLYNX = the cost is mostly due to the environment (power, network, etc.)<br>
&nbsp;<br>
&nbsp;<br>
Problem with starpoints:<br>
<ul>
    <li>
        Many people need the starpoint key for collaboration (EN/EL BE) but we need to keep control on who/when access the starpoint.</li>
    <li>
        Today, many keys were distributed and we don&rsquo;t know who is doing what.</li>
    <li>
        The risk is that someone damage a starpoint, with direct cost on equipement (up 500 KCHF) plus the impact of the organization (service failure for a couple of days).</li>
</ul>
&nbsp;<br>
Sharing key for piquet:<br>
<ul>
    <li>
        Rui does not like this much. Each person should have a key.</li>
    <li>
        To grant access, then you grant access to a person, regardless of his/her key.</li>
    <li>
        It could be that the electronic key is in a cabinet where it is needed to badge to get access to it.</li>
</ul>
&nbsp;<br>
Role based authentication<br>
<ul>
    <li>
        Define some roles to get access + eventual other constraints.</li>
    <li>
        Owner of resources will then allow roles to access the resource.</li>
    <li>
        Who assign the role? For which duration ? Prevent some known person to get a role.</li>
    <li>
        Only the person on duty can get access to the resource = doesn&rsquo;t look a good idea / too complicate to manage all exceptions.</li>
    <li>
        Less than a year to implement a role based authentication is not realistic.</li>
    <li>
        It could be linked to ADAMS and the AIS role system. Explanation to users is also difficult.</li>
</ul>
&nbsp;<br>
Fran&ccedil;ois is not 100% sure the LOCKEN system can replace the locks for the substation.<br>
Issue : TSO changing the locks for a building in which there are a starpoint or AUGs.<br>
Traceability is not really efficient, one can open a door, let it open and many people enter.<br>
Fran&ccedil;ois has some concerns about the people (external to his group) who will maintain the list of technicians will do it properly and knowing the safety risks they exposed people by granting access.<br>
Management tools : a view to see who has access for the location manager, a view for the supervisor to know who has access to what, a view for the user to know to what he has access.<br>
&nbsp;<br>
Rui would like to stress the LOCKEN system to check that it is scalable at CERN level.<br>
Frederic and Rui to discuss how to put test in place.<br>
Tony offers to give a couple of keys to EN and BE to access the starpoint and so they have them for eventual future deployment on their side.