/[pdpsoft]/trunk/grid-mw-security/ees/thesis/introduction.tex
ViewVC logotype

Contents of /trunk/grid-mw-security/ees/thesis/introduction.tex

Parent Directory Parent Directory | Revision Log Revision Log


Revision 654 - (show annotations) (download) (as text)
Thu Aug 20 14:09:25 2009 UTC (12 years, 5 months ago) by aramv
File MIME type: text/x-latex
File size: 7938 byte(s)
Added a lot of beautification to grid.tex. Added grid job diagrams. Added some citations.
1 \chapter{Introduction}
2 \section{Nikhef}
3 Nikhef (\textit{Nationaal instituut voor subatomaire fysica}) is the Dutch National institute for subatomic physics.
4 It's a collaboration between \textit{Stichting voor Fundamenteel Onderzoek der Materie} (FOM), \textit{Universiteit van Amsterdam} (UvA), \textit{Vrije Universiteit Amsterdam} (VU), \textit{Radboud Universiteit Nijmegen} and the \textit{Universiteit Utrecht}.
5 The name was originally an acronym for \textit{Nationaal Instituut voor Kernfysica en Hoge Energie-Fysica}, (National institute for nuclear and high energy physics).
6 After closing down the linear electron accelerator in 1998 the research into experimental nuclear physics yielded, but the Nikhef name has been retained up to the present day.
7 \cite{nikhefwebsite:overnikhef}
8
9 These days Nikhef is involved in areas dealing with subatomic particles.
10 Most employees at Nikhef are involved with the ATLAS and LHCb projects related to the \textit{Large Hadron Collider} (LHC).
11 High energy physics experiments generate vast amounts of data, analysis of which requires equally vast amounts of computing power.
12 In the past supercomputers were used to provide this power, but in order to perform analysis on high-energy, high-frequency subatomic particle interactions required by the LHC experiments a new method of pooling computing resources was devised: Grid computing.
13 The goal of Grid computing is to utilize existing infrastructures equally well as future infrastructures.
14
15
16 %It owned the second installation of a computer system in the Netherlands (the first being owned by the CWI).
17 \textit{Electronics Technology} (ET) and \textit{Computer Technology} (CT) are smaller departements within Nikhef. The CT departement provides Nikhef's computing infrastructure.
18 The DataGrid group is an offshoot of the CT departement which develops on Grid infrastructure, policy and software.
19 % Astro-particle physics and theory
20
21 % Geschiedenis nikhef -> computing resources nodig -> grid
22
23 % Grootste gedeelte bij ATLAS, dan LHCb, electronica, Grid,
24 % Grid is uitvloeisel van de CT groep.
25 % Meer mensen betrokken bij het bouwen van detectoren etc, dan met Grid computing. Meer mensen bezig met electronica, mechanica en ontwerp-afdeling dan grid computing.
26 % Supercomputers, high energy, lots of interactions, high frequency interactions, lots of analysis methods.
27 % Aanleiding voor nieuwe manier van computing. Namelijk: grid computing.
28 % Werd voorzien door CERN in eind jaren 90 dat de computing resources niet voldoende zouden zijn. In het jaar 2000 kwam de realisatie dat men moest gaan investeren in Grid computing. Of clusters aan elkaar knopen aangezien 'iedereen' al clusters had.
29 % Bestaande infrastructuren evengoed gebruiken als nieuw te bouwen infrastructuren, bij voorkeur tegelijk.
30
31 % Astro-particle-physics is een belangrijk onderwerp.
32 % Dan is er ook een theorie afdeling.
33 % Gravitational wave analyse. Bezig met onderdelen die ervoor zorgen dat de metingen preciezer worden.
34
35 \section{Grid Computing}
36 Grid computing is a term that describes computing platforms that aim to provide ubiquitous computing resources.
37 Analogous to the power Grid, software should just be 'plugged in' and run.
38 It is easy to mistake Grid computing for Cloud computing: the key distinction to make is that while cloud computing completely abstracts the underlying infrastructure from the user's point of view (for example by using virtualization), Grid computing tries to optimize available resources by maintaining control over the lower layers of the architecture.
39 This enables users to, for example, devise their own strategies for replicating data (within their project's quota limits).
40 This is abstracted from the user in cloud computing environments.
41 Also, Grids are inherently multi-domain (meaning they are provided by multiple cooperating organisations) and open (meaning they make use of open standards).
42 (Incidentally, they also frequently employ Open Source software.)
43 CERN, one of the largest users of Grid technology, define Grid computing simply as: "\textit{A service for sharing computer power and data storage capacity over the Internet}".
44 % Bij grid heb je binnen de quota limitieten de replicatie tactieken zelf in de hand.
45 % Bij cloud wordt het gekozen voor je.
46 % Multi-domain
47 % Open standaarden
48 % Open source
49
50 % Computing resources
51 % Meer cluster architectuur
52 % Bepaalde typen resources
53 % Hoe groot het grid is (zie mail oscar)
54
55
56
57 BiG Grid (The Dutch e-science Grid) has between 4500 and 5000 computing cores (not including LISA, which has 3000 cores) and about 1.5 petabytes of storage (with 100 terabytes of pre-stage cache space). The capacity of available tape storage is about 3 petabytes.
58
59
60
61 EGEE (The European Grid initiative) has roughly 100.000 computing cores and 50 petabytes of storage (most of which is tape storage).
62
63 % BigGrid:
64 % #CPU: ~4500 a 5000 cores (exclusief LISA. Die heeft ruwweg 3000 cores).
65 % #disk: 1,5PB en 100TB pre-stage cache. Tape is ongeveer nog iets van 3PB
66
67 % EGEE:
68 % #CPU: ~100k cores
69 % #disk: ~50PB, waarvan het grootste deel tape is en nog meer onbereikbaar.
70
71 \section{Participating International Organisations}
72 Like supercomputers, Grids attract science.
73 This has lead to a community of Grid computing users which advance the Grid computing field on an international scale.
74
75 Some of the cooperating organisations within the Grid computing community are:
76
77 \begin{itemize}
78 \item BiG Grid, the Dutch e-science Grid. An example of a \textit{National Grid Initiative} (NGI), of which there are many.
79 \item The \textit{European Grid for E-sciencE} (EGEE). A leading body for NGI's. To be merged into the \textit{European Grid Initiative} (EGI).
80 \item The \textit{LHC Computing Grid} (LCG) is the Grid employed by CERN to store and analyze data generated by the LHC. Also a member of EGEE.
81 \item The \textit{Virtual Laboratory for e-Science} (VL-e). A separate entity that employs Grid infrastructure independently.
82 \end{itemize}
83
84 % EGEE -> EGI
85 % BiG Grid == NGI, zo zijn er velen
86 % LCG is lid van EGEE
87 % VLEmed geen binding met EGEE, maar kan afkijken ervan doordat andere soorgelijke projecten er wel aan meedoen.
88 % Community rond de infrastructuur.
89 % Hetzelfde gebeurt bij supercomputers. Trekt wetenschap aan. -> Internationaal aanzien.
90
91 % EGI wordt een sturend orgaan voor de NGI's, die alle resources zelf moeten supplyen.
92
93 \section{PDP group \& security}
94 The PDP group at Nikhef is associated with BiG Grid, the \textit{LHC Computing Grid} (LCG), \textit{Enabling Grids for E-sciencE} (EGEE) and the (planned) \textit{European Grid Initiative} (EGI).
95
96 % Application Domain Analyst
97 Within Nikhef, the PDP group concerns itself with policy and infrastructure decisions pertaining to authentication and authorisation for international Grid systems.
98 It facilitates the installation and maintenance of computing, storage and human resources.
99 It provides the Dutch national academic Grid and supercomputing CA, and also delivers software such as:
100 \begin{itemize}
101 \item Grid middleware tools (gLite stack)
102 \item Cluster management software (Quattor)
103 \end{itemize}
104
105 % Cluster beheer
106 % Software certification & integration
107
108 % Primair bezig met infrastructuur up & running houden.
109
110 The PDP group employs \textit{Application Domain Analysts} (ADA's) which try to bridge the gap between Grid technology and its users by developing software solutions and offering domain-specific knowledge to user groups.
111 %Human resource tasks involve
112
113 % Software: grid middleware (security) user tools, cluster management software (quattor)
114 % faciliteren en onderhouden van computing en storage resources, en human resources.
115 % software development.
116 % beleidskwesties op internationaal niveau.
117
118 % Houdt zich bezig met authenticatie & authorisatie, en beleidskwesties die daarmee samenhagnen. policy
119 % Bijv: draait nationale academische grid en supercomputing CA.
120 % Data pooling
121 % Kringen werkzaamheden
122

grid.support@nikhef.nl
ViewVC Help
Powered by ViewVC 1.1.28