14 Distributed Computing
621
Some interesting alternatives are starting to emerge, which avoid Identity Federation. One approach is to identify a single reliable provider for AA, for example
the OpenID service from ORCID 25 combined with an appropriate attribute server.
Another approach is for a project or experiment to deploy a single AA scheme
available to services for that project around the world. This can be accomplished in
a simple and economical manner by re-using the project’s personnel database and
exposing it through the Internet in a secure manner using an ldap server. 26 These
type of implementations of single-sign-on are ideal for service providers that serve
a single project. On the other hand, they do shift the diversity problem to the service
providers, which must locally adjust crucial attributes such as uid and gid in order
to avoid duplications in multi-project or multi-experiment environments.
14.3.2 Processing and Storage
Two configurations of CPU servers are usually deployed:
• “Batch workers”, which execute tasks (jobs) that don’t require user interaction
and are scheduled using a “Batch” system 27 or a more sophisticated resource
harvester, such as HTCondor. 28
• “Interactive nodes”, where users can connect (or “log-in”) and perform work that
requires interaction via a screen, keyboard and mouse.
Several configurations of Disk servers are usually deployed:
• “Network File servers”, based on industry-standard or widely used protocols,
such as NFS, 29 smb2 30 or http with WebDAV, 31 are used to provide “home” and
“project” directories holding text and binary files of relatively small size. These
directories appear in the CPU servers as “mount points” or “network shares” that
largely behave as a local, conventional disk resource. The POSIX input–output
25 Haak, Laurel L., et al. “ORCID: a system to uniquely identify researchers.” Learned Publishing
25.4 (2012): 259–264.
26 Johner, Heinz, et al. Understanding LDAP. Vol. 6. IBM, 1998.
27 See for example Yoo, Andy B., Morris A. Jette, and Mark Grondona. “Slurm: Simple linux
utility for resource management.” Workshop on Job Scheduling Strategies for Parallel Processing.
Springer, Berlin, Heidelberg, 2003.
28 Fajardo, E. M., et al. “How much higher can HTCondor fly?” Journal of Physics: Conference
Series. Vol. 664. No. 6. IOP Publishing, 2015.
29 Shepler, Spencer, et al. Network file system (NFS) version 4 protocol. No. RFC 3530. 2003.
30 French, Steven M., and Samba Team. “A New Network File System is Born: Comparison of
SMB2, CIFS and NFS.” Linux Symposium. sn, 2007.
31 Goland, Yaron, et al. HTTP Extensions for Distributed Authoring – WEBDAV. No. RFC 2518.
1999.
621
Some interesting alternatives are starting to emerge, which avoid Identity Federation. One approach is to identify a single reliable provider for AA, for example
the OpenID service from ORCID 25 combined with an appropriate attribute server.
Another approach is for a project or experiment to deploy a single AA scheme
available to services for that project around the world. This can be accomplished in
a simple and economical manner by re-using the project’s personnel database and
exposing it through the Internet in a secure manner using an ldap server. 26 These
type of implementations of single-sign-on are ideal for service providers that serve
a single project. On the other hand, they do shift the diversity problem to the service
providers, which must locally adjust crucial attributes such as uid and gid in order
to avoid duplications in multi-project or multi-experiment environments.
14.3.2 Processing and Storage
Two configurations of CPU servers are usually deployed:
• “Batch workers”, which execute tasks (jobs) that don’t require user interaction
and are scheduled using a “Batch” system 27 or a more sophisticated resource
harvester, such as HTCondor. 28
• “Interactive nodes”, where users can connect (or “log-in”) and perform work that
requires interaction via a screen, keyboard and mouse.
Several configurations of Disk servers are usually deployed:
• “Network File servers”, based on industry-standard or widely used protocols,
such as NFS, 29 smb2 30 or http with WebDAV, 31 are used to provide “home” and
“project” directories holding text and binary files of relatively small size. These
directories appear in the CPU servers as “mount points” or “network shares” that
largely behave as a local, conventional disk resource. The POSIX input–output
25 Haak, Laurel L., et al. “ORCID: a system to uniquely identify researchers.” Learned Publishing
25.4 (2012): 259–264.
26 Johner, Heinz, et al. Understanding LDAP. Vol. 6. IBM, 1998.
27 See for example Yoo, Andy B., Morris A. Jette, and Mark Grondona. “Slurm: Simple linux
utility for resource management.” Workshop on Job Scheduling Strategies for Parallel Processing.
Springer, Berlin, Heidelberg, 2003.
28 Fajardo, E. M., et al. “How much higher can HTCondor fly?” Journal of Physics: Conference
Series. Vol. 664. No. 6. IOP Publishing, 2015.
29 Shepler, Spencer, et al. Network file system (NFS) version 4 protocol. No. RFC 3530. 2003.
30 French, Steven M., and Samba Team. “A New Network File System is Born: Comparison of
SMB2, CIFS and NFS.” Linux Symposium. sn, 2007.
31 Goland, Yaron, et al. HTTP Extensions for Distributed Authoring – WEBDAV. No. RFC 2518.
1999.