Joining as a Federated Resource Centre

Introduction

A Resource Centre (RC) is the smallest resource administration domain in the EGI Federation. It can be either localised or geographically distributed and provides a minimum set of local or remote IT Services compliant with well-defined IT Capabilities (HTC, Cloud, Storage, etc.) necessary to make resources accessible to Users. EGI is a Resource Infrastructure federating RCs to constitute a homogeneous operational domain.

Registration and certification

In order to join the EGI Infrastructure, a RC needs to present the request to the Research Infrastructure Provider (RP) existing in its country. A RP is a legal organisation, part of the EGI Resource Infrastructure, responsible for managing and operating a number of operational services at national level, supporting EGI RCs and user communities. Please have a look at the Operations Start Guide to get familiar with the terms mentioned above, and to have a complete picture of the several actors participating in our landscape.

The RP operators are going to guide and support the RC during the registration and certification procedures.

Firstly, the RC will be asked to read, understand, and accept:

• the RCs Operational Level Agreement (OLA), an agreement made between the RC and its RP that defines the minimum set of operational services and the respective quality parameters that a Resource Centre is required to provide in EGI;
• the Security Policies defined in EGI to guarantee that all the security aspects with the service delivery are fulfilled and enforced.

Secondly, the RC should be registered in the EGI Configuration Database: the provided information, from the generic contacts and roles of people to the service endpoints details, is needed to trigger the daily operations of other services and activities provided by the EGI Infrastructure such as the Monitoring of the resources, the Accounting, the Support, and the Security activities.

Once the entry in the Configuration Database is complete, the RP changes the RC status from “Candidate” to “Uncertified”, and the certification procedure can start: it comprises a series of technical controls to verify that the provided services work according to the expectations defined in the RC OLA. Any identified issue is notified by the RP operators to the RC and investigated until its solution.

When all the certification controls are successfully passed, the RC status is changed to “Certified” meaning that the RC is included in the EGI production infrastructure and its resources can be consumed by the users of the infrastructure.

Resource Centres responsibilities

Incidents and service requests

Providing support is a fundamental part of the daily activity of a provider participating in a large research infrastructure such as EGI. The support is not only for the users accessing the resources but also for those who are involved in the management and oversight of the infrastructure.

As defined in the RC OLA, the RC will handle incidents and service requests registered as tickets in the EGI Helpdesk service, with the expectation to acknowledge and process any notified issue, within the agreed response time associated with the priority of the ticket.

The response time is defined by the Quality of Support levels, and for the RCs the level will be Medium, meaning that there will 4 priorities for the incidents (requiring for example up to 5 working days for the “less urgent” tickets and up to 1 working day for the “top priority” ones), while any service request will be processed as “less urgent” ticket.

Security topics

The security posture of the infrastructure is framed by the set of policies constituting the Security Policies. Those policies cover different complementary activities including the operation of services, the processing of personal data and the management of security incidents and vulnerabilities.

Dealing with security incidents

The Security Incident Response Policy aims at coordinating the incident response across the infrastructure, ensuring that that incidents are promptly reported, and that all incidents are investigated as fully as possible.

Security incidents are to be treated as serious matters and their investigation must be resourced appropriately.

Resources Centres must report suspected security incidents to their RP Security Officer and to EGI Computer Security Incident Response Team (CSIRT) within 4 hours of discovery. This initial step will start the coordination of the incident response as documented in the procedure SEC01 EGI CSIRT Security Incident Handling Procedure.

• This procedure has been implemented according to the Security Incident Response Policy, to minimise the impact of security incidents affecting the Resource Centres part of the infrastructure
• It covers guidance on how the incident response should be coordinated, describing the responsibilities of the various parties, and encourages post-mortem analysis and promotes cooperation between Resource Centres.

Handling of vulnerabilities

The handling of vulnerabilities is a very formal process involving many different entities.

Anyone can report a software vulnerability via a form or by email contacting the Software Vulnerability Group (SVG).

The report will trigger an assessment, following the SEC02 Software Vulnerability Issue Handling procedure, by the Software Vulnerability Group, of the risk level associated with this vulnerability in the context of the activities of the EGI Infrastructure.

Once a vulnerability has been identified as presenting a risk to the infrastructure, it will be decided if an advisory should be prepared and circulated to the security contacts of the sites. After an agreed period of time, and depending on their confidentiality, advisories are made public.

Vulnerabilities identified as critical are handled according to the procedure SEC03 EGI-CSIRT Critical Vulnerability Handling. When applicable, this usually involves developing a custom security monitoring probe created to identify on High Throughput Compute RCs if their resources are vulnerable to the vulnerability. The status is closely monitored by the security team and accessible to the affected RCs.

Using this information correlated with the one from Pakiti, the patch management service collecting information about the patches deployed at the various High Throughput Compute RCs, the Incident Response Task Force (IRTF) on duty Security Officer will open tickets against the impacted sites according to the WI07 Security Vulnerability Handling procedure.

The EGI Service Delivery and Information Security (SDIS) team of the EGI Foundation, formerly known as the EGI Operations team, will follow up with the resource provider to work on resolving the ticket. The first duty of the resource provider is to acknowledge the vulnerability and then work on a prompt resolution as suggested in the ticket. In case a satisfactory resolution is not reached in due time, or a sign of active progress on addressing the vulnerability is not visible, the specific Resource Centre may be suspended.

Serving user communities

Once part of the production infrastructure, the RC is ready to deliver its resources to any of the users’ communities consuming the infrastructure.

The RC can continue serving local user communities, and at the same time deliver capacity for international user communities that approach EGI and therefore reach the federated RCs.

International communities reach EGI through the following channels:

• The EGI site where they can request access to services.
• The EGI-ACE Call for Use Cases.
• The Marketplace of the European Open Science Cloud.

Service and Operation Level Agreements (SLAs, OLAs)

The User Community Support team of the EGI Foundation receives these requests and negotiates the details of access, with the involvement of relevant and ‘fit-for-purpose’ RCs. This Service Level Management (SLM) process intervenes as a matchmaker between service expectations and needs of the user communities, acting as ‘customers’, and the capabilities of the RCs. Customers are enabled access to the resources in the form of Virtual Organisations (VOs).

In order to select providers for provisioning services to a given customer, technical requirements are collected from the customer then transferred to relevant providers. The Expression of Interests for support (EoIs) are collected from the interested providers during the negotiation phase, resulting in the best match with customer’s requirements and expectations (both technical and financial). Several aspects are considered during the negotiation phase, including the geographical location of the customer, national roadmap and priority of the providers, and costs of the service provisioning in case of a pay-for-use model.

The result of the negotiation is a ‘Service Level Agreement’ (SLA), and several ‘Operation Level Agreements’ (OLA), one with each contributing provider. (See SLAs-OLAs examples.)

SLAs and OLAs are typically signed for at least 1 year, and are automatically renewed, as long as the provider(s) or the customer do not express a decision to terminate the Agreement at least a month before the expiration date.

Performance reports: enforcing OLAs

As defined in the RCs OLA, the performance of the delivered services should meet the Service Level Targets: the monthly performance of the RCs is monitored, and when the targets are not achieved for three consecutive months, the affected RCs are notified through a ticket about the OLA violation and requested to provide within 10 working days an explanation for the low performance and a plan for improvement.

The RCs not providing a satisfactory explanation or not replying at all are eligible for suspension.

In order to re-join the EGI Infrastructure, any suspended RC should undergo a new certification procedure. The “Suspended” status cannot last for more than 4 months, after which a RCs is either in production again or definitely closed.

Besides the Targets defined in the RCs OLA, which are enforced to guarantee the permanence of the RC in the infrastructure, the targets promised to the users in the VO SLAs should also be met on a monthly basis: when a violation occurs, the RC is requested to provide a justification and a plan for improving the quality of the provided services.

If repeated violations occur, the SLA can be renegotiated with the customer, either by changing the Service Level Targets, or by choosing a different RCs as a provider.