DRAFT

WLCG Critical Services (Dev page!)

Introduction

This page lists per LHC experiment the set of services that are:

• not operated by its own personnel, and
• deemed critical for the successful operation of
  its grid workflows and for related activities.

Most of those services are hosted and operated by CERN-IT, while several Tier-1 sites and other partners also provide some.

For every relevant service, each experiment has provided indications of the effects of the service being unavailable. The impact indicates the effect on operations or people if the service were unavailable for a few days. The urgency indicates how quickly that impact would be reached. The criticality is defined as the product of urgency and impact. At the right hand side there are columns for the maximum criticality of a service across the experiments, the sum of the criticalities across the experiments and the weighted maximum criticality. The latter ranks services with identical maximum criticalities according to their respective sums of criticalities. Each numeric column can be sorted in ascending (descending) order by clicking once (twice) on its header.

Impact on operations and/or people

Urgency levels

Criticality levels

As a visual aid, 3 criticality ranges have been defined with distinct colors.
For a given experiment and for the maximum across the experiments, the ranges are as follows:

For the sum of the criticalities across the experiments:

The colors for the weighted maximum values correspond to those of the maximum values across the experiments.

Purpose of the tables

These tables are meant to clarify which services require which level of attention in their implementation and operation, to try and minimize the effects of service unavailability on the experiments, to the extent feasible. For example, a highly critical service should, if possible, be implemented and monitored in a more robust way than a less critical service. HA deployment methods, load-balancing and/or hot standby setups should be considered for such cases.

These tables do not make any promises about the level of support that can be expected for a given service: unless a specific arrangement was made for a particular service, the support level is best-effort for any service, though in practice it usually is compatible with the actual criticalities of the given service. If not, the service implementation and operation can be looked into.

Draft SNow support units

The (draft, suggested) SNow Functional Element (FE) or Service Element (SE) behind each service has now been added to the table.

For services that should be contacted through SNow instead of GGUS, the background of the FE or SE is set as in this

We will ask IT groups to validate these soon.

CERN-IT services

Services at other sites

Service		urg	imp	crit	urg	imp	crit	urg	imp	crit	urg	imp	crit		max	sum	wtd
		ALICE			ATLAS			CMS			LHCb				crit	crit	max
GOCDB		1	4	%x_alice%	4	4	%x_atlas%	4	4	%x_cms%	7	7	%x_lhcb%		%max%	%sum%	%wmax%
MyOSG				%x_alice%	4	4	%x_atlas%	4	4	%x_cms%			%x_lhcb%		%max%	%sum%	%wmax%
GGUS		1	4	%x_alice%	4	4	%x_atlas%	7	7	%x_cms%	7	4	%x_lhcb%		%max%	%sum%	%wmax%
FTS				%x_alice%	10	10	%x_atlas%	4	7	%x_cms%	4	10	%x_lhcb%		%max%	%sum%	%wmax%
Stratum-1		4	7	%x_alice%	7	4	%x_atlas%	4	7	%x_cms%	7	10	%x_lhcb%		%max%	%sum%	%wmax%
Accounting Portal		1	4	%x_alice%	1	4	%x_atlas%			%x_cms%	1	4	%x_lhcb%		%max%	%sum%	%wmax%

Previous versions

• Run 1: link
• Run 2: link
• LS2 until Oct 1st, 2020: link