ggus at peb –- 29.07.03 –- page 1 lcg klaus-peter mickel, gridka karlsruhe lcg-peb-meeting...
TRANSCRIPT
Klaus-Peter Mickel, GridKa Karlsruhe
GGUS at PEB –- 29.07.03 –- page 1
LCG
LCG-PEB-Meeting (29.07.03)
The Global Grid User Support Model (Report of GDB Working Group
5
and of related GDB discussions) 1. The history of the WG5 report
2. The Global Grid User Support Model
3. Procedures and Organization
4. Some screenshots from
www.ggus.org
Klaus-Peter MickelGridKa Karlsruhe(Forschungszentrum Karlsruhe, Germany)
Klaus-Peter Mickel, GridKa Karlsruhe
GGUS at PEB –- 29.07.03 –- page 2
LCG
The Global Grid User Support Model 1. The history of the WG5 report
Grid Deployment Board (GDB):09.12.02: Formation of WG5, mandate:
Recommend the initial LCG-1 user support model
06.02.03: Presentation of first ideas to the GDB21.02.03: WG5 Report v 1.0 is finished and in the web10.03.03: GDB: Considerable discussion on the scope of
the support model10.04.03: GDB: GridKa (Karlsruhe) will take
responsibility for the user support activity08.05.03: Some ideas concerning the GGUS web portal
presented to the GDB 23.05.03: WG5 Report v 2.0 is finished and in the web10.06.03: Online presentation of www.ggus.org to the
GDB 29.07.03: GGUS model presented to the PEB
Klaus-Peter Mickel, GridKa Karlsruhe
GGUS at PEB –- 29.07.03 –- page 3
LCG2. The Global Grid User Support Model
(GGUS Model - WG5) Members of the WG5:
Luca dell'Agnello (Italy)Ian Bird (Cern)John Gordon (UK) Thomas Kachelhoffer (France) Klaus-Peter Mickel (Germany) Laura Perini (Atlas)
People at GridKa (Karlsruhe), who have written the draft:Hans BaerRoland PietschmannWolfgang Thoene
The WG5 is completely convinced, that effective user services in a grid environment are absolutely essential for a greater success of the Grid.
Klaus-Peter Mickel, GridKa Karlsruhe
GGUS at PEB –- 29.07.03 –- page 4
LCG
WG5 – The Mandate
Recommend the initial LCG-1 user support model,
including:
Define the scope of responsibilities for a call center and/or
helpdesk type facility serving LCG;
Define the call center – one place or many ? – and define the
process for communication to ensure problem resolution
tracking;
Define acceptable user expectations (or SLA, service level
agreements), including the framework within which to
implement these;
Define how such a facility interacts with local support services
at sites providing grid resources;
Recommend procedures and tools to track and coordinate
distributed/grid problem resolution.
Klaus-Peter Mickel, GridKa Karlsruhe
GGUS at PEB –- 29.07.03 –- page 5
LCG
WG5 - GGUS – General Approach
We need a support for 24 hours on all days of the year The WG5 has proposed three different support centers in three
different time zones The WG5 has proposed one single point of access for all global
distributed users: One web portal (e.g. www.ggus.org) One mail address (e.g. [email protected])
should be switched to the active support center at a time
One phone number (e.g. +41-22-76-12345) should be switched to the active support center at
a time Communication between these three support centers should be
organized with a ticketing system via XML and with replicated databases
Klaus-Peter Mickel, GridKa Karlsruhe
GGUS at PEB –- 29.07.03 –- page 6
LCG
GGUS – General Approach cont'd
But: Many people say that 90 percent of all reported
problems are experiment specific and not Grid related and that the support people at the Global Grid User Support Centers would not be able to handle these experiment specific problems in a right and quick way.
The GDB has proposed, that the first line of support should be at experiment specialists, and that the Global Grid User Support should only be the second line of support.
But nevertheless: We need a Global Grid Support for 24 hours on all days of the year;
And therefore:The WG5 has proposed at least three different central support centers in three different time zones.
Klaus-Peter Mickel, GridKa Karlsruhe
GGUS at PEB –- 29.07.03 –- page 7
LCG GGUS – General Approach (cont'd)
The Grid Support Center on duty as the single point of access for all level 2 problems
(since the first level support should be gi-ven by the local experiment specialists)
Access via an interface to e.g. Savannah or via the web
Three different support centers in three different time zones
Europe8 - 16 h GMT
US16 - 24 h GMT
Far East0 - 8 h GMT
Local Grid operations
Local Grid operations
Local Grid operations
Local Grid operations
Local Grid operations
Local Grid operations
Customer/Experiment Level
Support Level Local operations Level
Klaus-Peter Mickel, GridKa Karlsruhe
GGUS at PEB –- 29.07.03 –- page 8
LCG
Local Operations Level: At Central Grid Operation Center and at each T-1-C (and also at each T-2-C?) e.g.:
• Problem solving• Maintenance• Local services• Resource management• Preventive activities• Problem announcements
The Support Model — three levels
Customer/Experi- Problem oriented Information orientedment Level: • Submit a problem • Ask for current Grid status,
• Track a problem • documentation, training
Support Level: At least three identical support centerswith: • Helpdesk application
• User, ticket and resource data base• Knowledge base
On call service outside the working hours
Klaus-Peter Mickel, GridKa Karlsruhe
GGUS at PEB –- 29.07.03 –- page 9
LCG 3. GGUS – Procedures
1. An user sends a request to "his" experiment specialists (on every possible way (web, mail, phone)) We need four experiment specialists' teams at each T-1-C !
2. The request is filed and numbered within the experiment specific bug tracking system (e.g. Savannah). This ticket belongs to the experiment's tracking system.
3. Someone at the experiments support level decides whether it is a Grid problem or not.
4. In case it is a Grid problem, the ticket is assigned to the GGUS team.
5. This assignment should be done using a XML interface or a “formatted” email to a central GGUS account.
6. At least the following data should be interchanged:User / Experiment / Problem description / Actions so far
Klaus-Peter Mickel, GridKa Karlsruhe
GGUS at PEB –- 29.07.03 –- page 10
LCG GGUS – Procedures (cont'd)
7. For Grid related problems GGUS will create a ticket within its own ticketing application and resend the ticket information to the Savannah system (using a XML message or an “formatted” email) with its own ticket number, and send an email to the user telling him his ticket number and how he can track this ticket. This ticket belongs to the GGUS ticketing system.
8. GGUS will work on the problem and whenever there is an update, an email is sent to the user and the Savannah ticket is updated using XML or “formatted” email
9. When the problem is solved a final email is sent to the user and a final update is sent to Savannah
10. If the user complains, asks or has additional questions or needs further information, he can use the GGUS web portal to look for his ticket
Klaus-Peter Mickel, GridKa Karlsruhe
GGUS at PEB –- 29.07.03 –- page 11
LCGOrganization of the GGUS
Centers There is one overall Grid Portal for all level 2 problems. Behind it are hidden at least three real support centers with well
trained support people for all grid related problems. The (at least) three real support centers should use the same software
and hardware systems. At all real support centers, there must be the identical (replicated)
databases and Grid user directories. All systems and support people are tripled; therefore a very
high reliability is reachable. As a ticketing system the GridKa people have evaluated and proposed
the commercial product Remedy (because it is well known, wide spread and reliable).
One should establish a knowledge base with all solutions of all user problems up to now. With such a knowledge base it should be possible to solve about 50 % of all problems "out of the box".
Klaus-Peter Mickel, GridKa Karlsruhe
GGUS at PEB –- 29.07.03 –- page 12
LCGThe Global Grid User Support Model
4. Some screenshots from
www.ggus.org
Remarks: - GridKa has reserved this domain for LCG
- The Grid portal www.ggus.org is working at present in an early test version
- The German HEP users at GridKa are beginning to test this GGUS system; for that we have established eight experiment specialists' mailing lists - one for each GridKa HEP experiment
- The authorization to use this portal from everywhere should be controlled later on by the central user administration (maybe a LCG wide LDAP directory structure)
Klaus-Peter Mickel, GridKa Karlsruhe
GGUS at PEB –- 29.07.03 –- page 13
LCG
Klaus-Peter Mickel, GridKa Karlsruhe
GGUS at PEB –- 29.07.03 –- page 14
LCG
Klaus-Peter Mickel, GridKa Karlsruhe
GGUS at PEB –- 29.07.03 –- page 15
LCG
Klaus-Peter Mickel, GridKa Karlsruhe
GGUS at PEB –- 29.07.03 –- page 16
LCG
Klaus-Peter Mickel, GridKa Karlsruhe
GGUS at PEB –- 29.07.03 –- page 17
LCG
Klaus-Peter Mickel, GridKa Karlsruhe
GGUS at PEB –- 29.07.03 –- page 19
LCG
Klaus-Peter Mickel, GridKa Karlsruhe
GGUS at PEB –- 29.07.03 –- page 20
LCG
Klaus-Peter Mickel, GridKa Karlsruhe
GGUS at PEB –- 29.07.03 –- page 21
LCG
Klaus-Peter Mickel, GridKa Karlsruhe
GGUS at PEB –- 29.07.03 –- page 22
LCG