Abstract

Today the model of computing assistance is no longer the one of few years ago. Members of our user community may be confused by the subtle but unavoidable changes which make a Helpdesk and User Support in general remain efficient.

This article outlines a brief history and the present day working strategy of the Central Computing Helpdesk. It seems that many members of the user community have a misconception of the Helpdesk model used at CERN today, and this article should clarify any doubts on the working environment of the Desk.

Helpdesk assistance in perspective

The idea of 'User-Support' in computing (first-level assistance) has been at CERN since the computers were first introduced. The service has been named by a range of well known acronyms such as the PEO ("Program Enquiry Office"), the UCO ("User Consultancy Office"), etc.

The original conception of the Helpdesk was that of a first level support entry point, where a skilled computing "generalist" (CERN staff) was available on shift to answer technical questions and ready to troubleshoot/debug difficult programming problems. This conception worked well in the days of mainframes, a contained number of operating systems (CDC/NOS-BE, IBM/MVS, IBM/VM and VAX/VMS) and Fortran as a unique programming language.

The rundown of CERNVM in June 1996 was a milestone. A new computing paradigm (that of Distributed Computing) was already taking over. The scientific user community, receiving till then a well focused computing support and service (one computer, one operating system, one single point of failure) migrated to distributed UNIX clusters in order to cope with the ever-increasing expectations for further computing capacity. The computing environment became understandably more complex; so did the support to users.

Today 'casual' CERN computing users pose questions and are confronted with difficulties in using mail clients, Web browsers and other "productivity" applications; on PCs, Macintosh as well as the central Unix clusters. Add to that the 'experienced' user who uses many programming languages, C, C++, Fortran, Java, etc., and a wide range of platforms (HP, Sun, IBM, Linux, Windows 95 and NT, etc.).

You may call it a support nightmare.

Paradigms also apply to Helpdesking

The contrast between the relative simplicity of the basic support model in the 80's and now is worth mentioning. First and foremost we have many more questions on many more topics. In addition the probability for any operational hardware failure likely affect a fraction of the user community is by nature higher. As are the expectations users have of computer stability to confront the current computational challenges. With this complex and also constantly changing environment a new type of helpdesk model has to exist.

First of all it has become increasingly important to reception all demands for intervention rather than concentrating in solving user problems one by one. The perception otherwise is of a support centre which "is never available when it is really needed". The calls are dispatched asynchronously to the area where it is most efficient to treat the problem.

Second, CERN has been analyzing areas which could be outsourced in a situation of shrinking resources. Operational activities are more suitable to be run via service contracts, where the quality of a rota system no longer relies as much on the particular skills of each individual, but rather on the integration of the members in a solid team.

In 1998 CERN decided on an out-sourced Computing Helpdesk and as of January 1999 a consortium of WM-data and DCS has been running the Computing Helpdesk in building 513 (see related article on "Service Handbook for the Desktop Support Contract").

One desk, one number, one e-mail

Already in October 1998 a new (easy-to-remember) phone number (78888) was introduced, in order to facilitate requests for interventions.

In January this year the desk has been moved to join up with the Operations staff in an effort to help communications in cases of hardware failure and optimise the space available. The move was prompted initially by the fact that the two staff were rather isolated and relied on information from others regarding hardware and software problems. Being next to the operators gives the Helpdesk staff a chance to interact and help each other trouble-shoot problems via user feedback. Another major factor for the move was to have one unique window for the user-interface, hence clarifying the entry point for all user problems in the Computer Center (building 513).

The helpdesk now works on a tiered system. Two persons staff the desk during prime shift (8:30 to 17:30, Monday to Friday). [Outside these hours calls are routed to the computer operator who will try to answer questions on a best effort basis or pass these questions on to the helpdesk proper.] The members of staff on duty receive requests via phone and e-mails as well as visits to the desk in building 513; they have a large knowledge database at their disposal as well as many tools to help with their work and try to answer as many questions as they can immediately.

Should the staff on duty not be able to reply to the question, or that it requires the assistance of a person directly on the machine, the problem is 'escalated' to another level. This support level might be another member of the out-sourced group for 'local support' or it might go to a second level of support within IT. Bearing this in mind, remember that when you report a problem to the helpdesk it is imperative to give them as much information as possible relevant to the problem, as it is not necessarily the person to whom you report the problem who will supply the answer. If your problem is escalated you will receive an informative message within 2 business hours. This message should contain a reference number (a so-called trouble "ticket"); you should note this number and quote it on all subsequent requests for information about your request.

Remember also that the Helpdesk 'hosts' support lines to filter simple problems; hence even though your problem might be sent to Mail.Support@cern.ch it will pass through the Helpdesk anyhow, such that straightforward problems are answered immediately. With this in mind the goal is to reduce these 'support' lines to a minimum such that they disappear and all problems are then routed centrally to and from the Helpdesk.

Another recent innovation at the helpdesk is the use of a problem report tracking system. The commercial system chosen by the Laboratory is the Remedy Action Request system and is used widely in IT and other divisions to track problems and provides statistical information. In the near future all mail received by the helpdesk will be automatically channelled into this system so that no mail should ever be lost.

Where and how do you get help today?

The 'Central Computing Helpdesk' operates during weekdays from 8.30 - 17.30 and users may call the central number

Or they may prefer to send electronic mail to

We invite users to retain this information wherever this might be relevant (e.g. near desktop computers) as it had been announced (with a pre-printed note) in a recent Weekly Bulletin.

For your information, the Helpdesk also receptions at present the following support lines:

Mail.Support@cern.ch,Nice.Support@cern.ch, WWW.Support@cern.ch (or Web.Support@cern.ch), Netscape.Support@cern.ch, EDMS.Support@cern.ch, CDD.Support@cern.ch

and more will soon be added.

We encourage all users to take advantage of the Central Computing Helpdesk, but also of the public knowledge database managed by the User Support group in IT, that is available at http://cern.ch/consult/xfind and where you might even find the answer to your question before sending it ! Note that a small fraction of this knowledge database (from the 'Question & Answer' segment) is already displayed in all CNL issues, in the regular section "Questions and Answers from the Computing Helpdesk". The complete list of 'Question & Answer' is available at http://cern.ch/consult/qas with a full text search based on keyword(s) input. There are regularly reviewed and user feedback on specific "qas" is welcome (because it is not precise enough, maybe incorrect or even out-of-date).