The SAMGrid Fabric Services

1
The SAM-Grid Fabric Services

• Gabriele Garzoglio (for the SAM-Grid team)
• Computing Division
• Fermilab

2
Overview

• Introduction
• The grid-level services an overview
• Job Management
• The fabric-level services
• Local batch system adaptation
• Dynamic product retrieval
• Local sandbox management
• Job complex-status logging

3
Introduction

• SAM is a Data Handling System for HEP the
  project was started in 1997 by DZero
• SAM-Grid project started in 2001-2002 to handle
  DZeros expanded needs for globally distributed
  computing
• CDF joined SAM-Grid at the end of 2002
• JIM complements the data handling system (SAM)
  with Job and Info ManagementSAM-Grid JIM
  SAM
• JIM is funded by PPDG and GridPP
• Participated at SC02 and SC03

4
Overview

• Introduction
• The grid-level services an overview
• Job Management
• The fabric-level services
• Local batch system adaptation
• Dynamic product retrieval
• Local sandbox management
• Job complex-status logging

5
Job Management
User Interface
User Interface
Submission Client
Submission Client
Match Making Service
Match Making Service
Broker
Queuing System
Queuing System
Information Collector
Information Collector
JOB
Data Handling System
Data Handling System
Data Handling System
Data Handling System
Execution Site 1
Execution Site n
Computing Element
Computing Element
Computing Element
Storage Element
Storage Element
Storage Element
Storage Element
Storage Element
Grid Sensors
Grid Sensors
Grid Sensors
Grid Sensors
Computing Element
6
Overview

• Introduction
• The grid-level services an overview
• Job Management
• The fabric-level services
• Local batch system adaptation
• Dynamic product retrieval
• Local sandbox management
• Job complex-status logging

7
Running jobs on Grid resources the trend

• Grid resources are not dedicated to a single
  experiment
• Translation
• no daemons running on the worker nodes of a Batch
  System
• no experiment specific software installed

8
Running jobs on Grid resources today

• The situation is transitioning
• Generally, experiments can install specific
  services on a node close to the cluster.
• Worker nodes typically access the software via
  shared FS not scalable!
• Local resource configuration still too diverse to
  easily plug into the Grid
• Today, most of our efforts are directed to coping
  with (the lack of) standard local fabric services

9
Overview

• Introduction
• The grid-level services an overview
• Job Management
• The fabric-level services
• Local batch system adaptation
• Dynamic product retrieval
• Local sandbox management
• Job complex-status logging

10
Motivation

• Problem standard grid batch system adapters
  (globus job-managers) are too restrictive to fit
  all the local configurations
• Examples
• the terms of the agreement for using the batch
  system can be expressed with special directives
  to the batch system
• system administrators end up writing wrappers
  around the standard batch system commands

11
SAM Batch System Adapter

• We factor out the local batch system
  configuration using an intermediate layer that
  abstracts the basic interactions with the batch
  system
• submit command
• lookup command
• remove command
• For each of the commands above, the administrator
  can specify how to parse the output to fish out
  the relevant information e.g. local job id when
  submitting
• We have written JIM globus job managers that use
  this layer

12
Overview

• Introduction
• The grid-level services an overview
• Job Management
• The fabric-level services
• Local batch system adaptation
• Dynamic product retrieval
• Local sandbox management
• Job complex-status logging

13
Motivation

• Portability of the software for DZero and CDF is
  still a problem not completely solved.
• Most of the CDF and DZero applications still rely
  on the offline software to be preinstalled at the
  site.
• Administrators need to install and maintain the
  software at each site
• A job submitted to the grid must be able to
  execute at a site where its dependencies are
  installed

14
Old solution software advertisement

• Administrators install the software at each site
• The JIM advertisement framework senses the new
  product and advertises it to the broker as one of
  the characteristics of the site
• Drawbacks
• the administrators still need to install the
  software
• increased complexity of the advertisement
  framework it needs to know how to detect the
  list of installed products
• increased complexity of the broker it needs to
  enforce the matching to the eligible sites
• jobs running on old software versions may not
  find an eligible site

15
New solution dynamic software retrieval

• Product developers store the software into SAM
  with appropriate metadata
• Before running a job at a site, the
  infrastructure asks SAM for the delivery of the
  dependent products
• The products live in the SAM cache and are
  automatically managed
• Drawbacks
• increased complexity of local job submission

16
Overview

• Introduction
• The grid-level services an overview
• Job Management
• The fabric-level services
• Local batch system adaptation
• Dynamic product retrieval
• Local sandbox management
• Job complex-status logging

17
Nomenclature

• Input sandbox
• from the client (user sandbox)
• the executable
• configuration files
• special dependencies (libraries, products,)
• from the local site
• the product dependencies
• Output sandbox
• stdout, stderr
• log files
• small custom output (e.g. histograms)

18
Requirements

• We want an infrastructure that
• Locally stores the user sandbox (from the Grid)
  at the site
• transports and installs the input sandbox to the
  worker node
• packages the output and hands it over to the Grid

19
Limitations to overcome

• the file transport mechanism of a batch system is
  site specific and needs to be factored out
• shared file systems have scalability limits we
  want to rely on them as little as possible
• the worker nodes may have connectivity
  restrictions (firewalls)

20
The sandbox management 1

• It creates a sandbox area (reorganizing the
  native globus gass cache)
• It starts up a gridftp server for the
  communications between worker nodes and head node
  (no shared FS)
• It requests the delivery of the product
  dependencies
• It creates a self extracting archive that
  contains the gridftp client and a bootstrapping
  script when running, this transfers and installs
  the product dependencies, then passes control to
  the application

21
The sandbox management 2

• It submits to the batch system parallel instances
  of the self extracting archive
• The job relies on SAM for large input/output
  files transfers
• When the job finishes, stdout/stderr custom
  output is packaged at the head node to be
  transported back to the submission site via grid
  mechanisms

22
Open problems

• Not all the batch system allow the selection of a
  node with sufficient scratch space to install the
  needed software
• We would greatly simplify this infrastructure if
  there were a standard local storage service at
  all the sites (e.g. DiskFarm)

23
Overview

• Introduction
• The grid-level services an overview
• Job Management
• The fabric-level services
• Local batch system adaptation
• Dynamic product retrieval
• Local sandbox management
• Job complex-status logging

24
Motivation

• Distributed logging of job status/history
• Web monitoring
• Statistics on historical data
• Grid scheduling based upon job status/history at
  a certain site

25
The XML DB Status Logger

• The status of the job is reported to an XML
  database deployed at each execution site
• The information comes from the local batch system
  (simple job status e.g. idle, running, ) AND
  from the application (complex status e.g.
  Processing executable X in the chain)
• The XML database gives flexible remote access via
  standard mechanisms, such as XPath

26
Conclusions

• The SAM-Grid offers an extensible working
  framework for Grid-level Job/Data/Info Management
• The SAM-Grid adopts Fabric-level configurable
  solutions for batch system adaptation, product
  delivery, sandboxing and job complex-status
  logging
• The community needs to come up with standard
  fabric-level services to make any Grid usable

27
More info at

• http//www-d0.fnal.gov/computing/grid/
• http//samgrid.fnal.gov8080/
• Morag Burgon-Lyons Talk on SAM-Grid for CDF!