ReSS: A Resource Selection Service for the Open Science Grid Page: 4 of 10
This article is part of the collection entitled: Office of Scientific & Technical Information Technical Reports and was provided to UNT Digital Library by the UNT Libraries Government Documents Department.
Extracted Text
The following text was automatically extracted from the image on this page using optical character recognition software:
2.1 Glue Schema to old classad mapping
The Glue Schema is a resource description model adopted by commercial companies
and Grid organizations. Resources are described in terms of entities, such as Clusters
or Storage Areas, and their logical relationships, like association or aggregation, are
expressed using UML diagrams.
In order to use readily available match-making services for resource selection, ReSS
describes resources in old classad format, an unstructured list of [attribute , value]
pairs. The project therefore faced the problem of mapping the Glue Schema structure
into a set of unstructured classads. The activity was conducted in collaboration with
the Glue Schema group and resulted in a mapping document [7] and its
implementation in CEMon.
In the Glue Schema, a computational resource is described by a "Cluster", which is
part of a computing center or "Site". A Cluster is composed by one or more groups
("Subclusters") of homogeneous computing nodes ("Hosts"). Hosts are characterized
by parameters related to the processor type, memory, operating system, etc. Access to
the Cluster can be achieved by one or more gateways or queues ("Computing
Elements" or CE). The Computing Element is described by informational parameters,
such as the gateway address, state parameters, such as the number of running or idle
jobs, policy parameters, such as the maximum wall clock time allowed by the local
scheduler, etc. In addition to total values for these parameters, like the total number of
running jobs at the CE, the model also allows for VO-specific values ("VOView"),
like the number of running jobs for a specific VO.Fig. 2. Mapping the Glue Schema "tree" (left) into a set of "flat" classads (right): the algorithm
navigates through all possible combinations of (Cluster, Subcluster, CE, VO)
Figure 2 shows a schematic UML representation of a computational resource. The
mapping between this structure and a set of old classads is built considering all
possible combinations of inter-related CE, Cluster, and Subclustrer entities. In other
words, each combination (classad) contains a single CE, Cluster, and Subcluster
entity. In addition, if VO-specific parameters are available to characterize the CE (VO
View entity), these are used instead of the general CE attributes.--SubCluster1
-SubCluster2
Site ->Cluster
-- r-'vo1
CE1
V02
vo2
CE2 +
V03Site "
Cluster Site Is
SubCluster1 Cluster
CE1 'SubCluster2
V01 CE1
r=--I vo1
Site
Cluster
SubCluster1
vo2
Upcoming Pages
Here’s what’s next.
Search Inside
This article can be searched. Note: Results may vary based on the legibility of text within the document.
Tools / Downloads
Get a copy of this page or view the extracted text.
Citing and Sharing
Basic information for referencing this web page. We also provide extended guidance on usage rights, references, copying or embedding.
Reference the current page of this Article.
Garzoglio, Gabriele; Levshina, Tanya; Mhashilkar, Parag & Timm, Steve. ReSS: A Resource Selection Service for the Open Science Grid, article, January 1, 2008; Batavia, Illinois. (https://digital.library.unt.edu/ark:/67531/metadc895957/m1/4/: accessed March 28, 2024), University of North Texas Libraries, UNT Digital Library, https://digital.library.unt.edu; crediting UNT Libraries Government Documents Department.