The Grid Middleware (Fundamentals of Grid Computing)

1
The Grid Middleware (Fundamentals of Grid
Computing)

• Adam Belloum
• Computer Architecture Parallel Systems group
• University of Amsterdam
• adam_at_science.uva.nl

2
What did we learn in the introductory course?

• Introduction to Grid computing

3
What the Grid Can do

• Exploring underutilized resources
• Parallel CPU Capacity
• Virtual resources and Virtual Organization
• Access to additional resources
• Resource Balancing
• Reliability
• Management

4
The Best of Two Worlds
Open Grid Services Architecture
manage
share
access
Resources on demand
Applications on demand
Global Accessibility
Secure and universal access
Business integration
Vast resource scalability
Grid Protocols
Web Services

• Open Grid Services Architecture Evolution, J.P.
  Prost, IBM Montpellier, France, Ecole Bruide 2004

5
Grid and Web Services Convergence?
Grid
Web
However, despite enthusiasm for OGSI, adoption
within Web community turned out to be problematic
WS-Resource Framework Globus Alliance
Perspectives I. Foster
6
Grid and Web ServicesConvergence Yes!
Grid
Web
The definition of WSRF means that Grid and Web
communities can move forward on a common base
WS-Resource Framework Globus Alliance
Perspectives I. Foster
7
Summary

• Initial exploration (1996-1999 GT 1.0)
• Extensive application experiments core protocols
• Data Grids (1999-2002 GT 2.0)
• Large-scale data management and analysis
• Open Grid Services Architecture (2001-2003, GT
  3.0)
• Integration w/ Web services, hosting envs, res
  virtualization
• Databases, higher-level services
• Radically scalable systems (2003-??)
• Sensors, wireless, ubiquitous computing

8
Outline

• Type of Grid Resources
• Scheduling, reservation, and scavenging
• Grid Construction
• Grid Middleware components
• Deploying the Grid Software
• Configuring the Grid Middleware

9
Grid resources Computation

• CPU Cycles. There are 3 primary ways to exploit
  the computation resources of a grid.
• Execute an existing application on a grid host
• Parallel execution on several grid hosts
• Execute an application as many times on many grid
  hosts
• Scalability is a measure of how efficiently the
  multiple processors on a grid are used.

10
Grid resources Storage

• A grid provides an integrated view of data
  storage
• Machine on the grid provides a quantity of
  storage
• Temporary storage, attached to the Processor
  fast, but volatile,
• Secondary storage increases capacity,
  performance, sharing, and reliability of data
• Capacity can be increased by using the storage on
  multiple machines with a unifying file system.
• Data sets can span several storage machines,
  eliminating size restrictions often imposed by
  file systems

11
Grid resources Storage

• Advanced file systems on a grid can
• automatically duplicate sets of data, to provide
  redundancy for increased reliability and
  increased performance.
• Intelligent scheduler can select the storage
  location, based on usage pattern

12
Grid resources Communication

• Communications are important for
• sending jobs and their required data to points
  within the grid.

• Communication Bandwidth
• Is the most important feature needed to setup an
  effective grid.
• Redundant communications are sometimes needed to
• better handle potential network failures and
  excessive data traffic.

13
Grid resources Software licenses

• The grid may have software installed that may be
  too expensive to install on every grid machine.
• Using a grid, the jobs requiring this software
  are sent to the particular machines on which this
  software happens to be installed

14
Grid Resources Special Equipments

• Platforms on the grid often have different
  architectures, operating systems, capacities, and
  equipment.
• Each of these attributes represents a different
  kind of resource that the grid can use as
  criteria for assigning jobs to machines
• Such attributes must be considered when assigning
  jobs to resources in the grid

15
How to use the Grid Jobs Applications

• Grid resources are accessed executing
  application or job.
• The term application as the highest level of
  work on the grid.
• Applications may be broken down into any number
  of individual jobs.
• Those, in turn, can be further broken down into
  sub-jobs.
• Jobs are programs that are executed at an
  appropriate point on the grid. They may
• compute something,
• execute one or more system commands,
• move or collect data,
• or operate machinery.

16
How to use the Grid Jobs Applications

• A grid application that is organized as a
  collection of jobs is usually designed to
• have the jobs execute in parallel on different
  machines in the grid.
• The jobs may have specific dependencies that may
  prevent them from executing in parallel in all
  cases.
• Jobs may require some specific input data that
  must be copied to the machine on which the job is
  to run.
• Jobs may require the output produced by certain
  other jobs and cannot be executed until those
  prerequisite jobs have completed executing.

17
How to use the Grid Jobs Applications

• Jobs may spawn additional sub-jobs, depending on
  the data they process.
• This workflow can create a hierarchy of jobs and
  sub-jobs.
• The results of the jobs must be collected and
  assembled to produce the ultimate answer for the
  application

18
Resource Management

• Scheduling, Reservation, Scavenging

19
Scheduling, Reservation, Scavenging

• In the simplest of grid systems
• User selects a machine for running his job and
  then executes a grid command that sends the job
  to the selected machine.
• More advanced grid systems would include a job
  scheduler
• that automatically finds the most appropriate
  machine.

20
Scheduling, Reservation, Scavenging

• In a scavenging grid system
• Any machine that becomes idle reports its idle
  status to the grid management node.
• The management node assigns to the idle machine
  the next job that is satisfied by the machines
  resources.
• Scavenging is usually implemented in a way that
  is unobtrusive to the normal machine user.
• If the machine becomes busy with local non-grid
  work, the grid job is usually suspended or
  delayed.

21
Scheduling, Reservation, Scavenging

• Grid resources can be reserved in advance for a
  designated set of jobs.
• This is done to meet deadlines and guarantee
  quality of service.
• When policies permit, resources reserved in
  advance could also be scavenged to run lower
  priority jobs when they are not busy during a
  reservation period.
• Various combinations of scheduling, reservation,
  and scavenging can be used to more completely
  utilize the grid.

22
Grid construction
23
Deployment planning

• Security
• It is important to understand exactly which
  components of the grid must be rigorously secured
  to deter any kind of attack.
• It is important to understand the issues involved
  in authenticating users and properly executing
  the responsibilities of a CA.
• For example in an organization
• It is important to understand how the departments
  in an organization interact, operate, and
  contribute to the whole

24
Grids are by definition Heterogeneous

• Grid is about legacy resources, infrastructure,
  applications, policies, and procedures
• The grid and its administrators must integrate in
  stealth modewith
• Firewalls
• Filesystems
• Queuing systems
• Grumpy systems administrators
• Tried and true applications

25
Challenges in Grid Computing

• Reliable performance
• Trust relationships between multiple security
  domains
• Deployment and maintenance of grid middleware
  across hundreds or thousands of nodes
• Access to data across WANs
• Access to state information of remote processes
• Workflow / dependency management
• Distributed software and license management
• Accounting and billing

26
Grid Middleware Components

• Submission software
• Machine of a grid can be used to submit jobs
  initiate grid queries.
• Some grid systems, use a separate component
  installed on submission nodes.
• Management software
• Large grids have a hierarchical matching the
  connectivity topology.
• machines locally connected together with a LAN
  form a cluster
• Scheduling Software
• round-robin fashion
• implement a job priority system.

27
Grid software components

• Communications
• A grid system may include software to help jobs
  communicate.
• Open standard MPI is often included as part of
  the grid system
• Observation, management, and measurement
• The donor software include some tools that
  measure the current load on a machine using
  either OS tools or by direct measurement.
• It is sometimes referred to as a load sensor.
  Some grid systems provide the means for
  implementing custom load sensors for other than
  CPU or storage resources.

28
Interacting/Using with the Grid

• User Perspective
• Application developer Perspective
• Administrator Perspective

29
What the Grid Means To

• The end user
• Can transparently access resources in multiple
  VOs
• Can more easily collaborate with other
  researchers
• The IT administrator
• Has a secure framework for implementing
  distributed resource sharing
• Local resource administrators can control access
  to their resources
• The manager
• Sees better utilization of capital resources
• Has a tool that helps break down organizational
  barriers

30
Using a grid A users perspective

• Enrolling and installing grid software
• A user enrolls as a grid user and installs the
  provided grid software on his own machine.
• He may optionally enroll his machine as a donor.
• Enrolling in the grid requires authentication
• certificate authority
• Installing the grid software on a machine
• for the purposes of using the grid
• as well as donating to the grid.

31
Using a grid A users perspective

• For users, the primary requirement is
  simplicity Access to the virtual organizations
  resources should not be significantly different
  from access to the local organizations
  resources.
• There should be a single sign-on, where users
  need to log on only once to access all permitted
  resources.
• Programs running on a users behalf should
  possess a subset of the users rights and have
  access to the permitted resources.

Protected channel
passwd
Randy Butler et al. A Natioanal-Scale
Authentication Infrastructure
32
Using a grid A users perspective

• Logging onto the Grid
• To use the grid, most grid systems require the
  user to log on to a system using a user ID that
  is enrolled in the grid
• Globus, for example, implements a proxy login
  model that keeps the user logged in for a
  specified amount of time, even if he logs off and
  back on the operating system and even if the
  machine is rebooted.

33
Using a grid A users perspective

• Queries and submitting jobs
• Grid systems provide command line tools as well
  as GUIs for queries.
• Job submission consists of 3 parts, even if there
  is only one command required.
• Input data and possibly the executable program or
  script are sent to the machine to execute the
  job.
• Job is executed on the grid (inside a protective
  sandbox)
• Results of the job are sent back to the submitter.

(1)
(3) Monitor/Control
(2) Configuration
34
Using a grid A users perspective

• Data configuration
• The data accessed by the grid jobs may simply be
  staged in and out by the grid system
• a large amount of data traffic
• There are many considerations in efficiently
  planning the distribution and sharing of data
• Data replication and reallocation

35
Using a grid A users perspective

• Monitoring progress and recovery
• The user can query the grid system to see how his
  application and its sub-jobs are progressing
• A grid system, in conjunction with its job
  scheduler, often provides a recovery for
  sub-jobs that fail.
• Programming error
• Hardware or power failure
• Communications interruption
• Excessive slowness

36
Using a grid A users perspective

• Reserving Resources
• User may arrange to reserve a set of resources in
  advance for his exclusive or high priority use.
• Planned Hardware
• Software maintenance events
• In a scavenging grid, it may not be possible to
  reserve specific machines in advance.
• Grid management systems may allocate a larger
  fraction of its capacity for a given reservation
  to allow for the likelihood of some of the
  resources becoming unavailable.

37
Using a grid Admins Perspective

• The concerns of resource-providing sites
  constrain an authentication and authorization
  infrastructure in two ways
• have there local (intra-domain) security policy
• typically cannot easily replace or modify their
  intra-domain security solution.
• How they see inter-domain security
• a distinct inter-domain solution that
  interoperates with local security solutions and
  is at least as strong as local solutions (will
  not weaken site security), is easy to understand
  (administrators can trust it).
• Administrators must have tight control over
  policies governing access to their resources,

Randy Butler et al. A Natioanal-Scale
Authentication Infrastructure
38
Using a grid Admins Perspective

• Planning
• Administrator should understand the
  organizations requirements for the grid to
  better choose the grid technologies that satisfy
  those requirements.
• Installation
• Selected grid system must be installed on an
  appropriately configured set of machines.
• Of prime importance is understanding the
  fail-over scenarios for the given grid system

39
Using a grid Admins Perspective

• Root access to both the node managing the grid
  and the donor machine
• Software to be installed on the donor machines
  need to be customized so that it can find the
  grid management machines automatically.
• Application software and data that should be
  installed on donor machines as well.
• This software may have specific licensing
  restrictions that should be understood and
  adhered to.

40
Using a grid Admins Perspective

• Managing enrollment of donors and users
• Machines donating resources Donor machines may
  have access rights
• Users controlling the rights of the users in the
  grid.
• As users join the grid,
• identities must be positively established and
  entered in the CA.
• In some cases, the administrator must propagate
  the user information to several or all grid
  machines.
• A new Donator machine has be added to the grid,

41
Using a grid Admins Perspective

• Certification authority
• An important aspect of maintaining strong grid
  security.
• You may choose to use an external CA or operate
  one.
• You must be able to trust the CA to strictly
  adhere to its responsibilities.
• The responsibilities of a certificate authority
  are
• Serve signed certificates to those needing to
  authenticate entities
• Maintaining a namespace of unique names for
  certificate owners
• Positively identify entities requesting
  certificates
• Issuing, removing, and archiving certificates
• Protecting the certificate authority server
• Logging activity

42
Using a grid Admins Perspective

• Resource Management this includes setting
  permissions for grid users
• use the resources
• track resource usage
• implementing a corresponding accounting or
  billing system.

43
Using a grid Admins Perspective

• Data sharing
• For small grids, the sharing of data can be
  fairly easy, using existing networked file
  systems, databases, or standard data transfer
  protocols.
• As a grid grows and the users become dependent
  on any of the data storage repositories
• procedures to maintain backup copies and replicas
  to improve performance.

Requested Data Management System ?
Services specific to the data Grid
infrastructure
Low level Services (shared with other Grid
Components)
44
Using a grid An application developers
perspective

• Grid applications can be categorized in one of
  the following three categories
• Applications that are not enabled for using
  multiple processors but can be executed on
  different machines.
• Applications that are already designed to use
  the multiple processors of a grid setting.
• Applications that need to be modified or
  rewritten to exploit a grid.
• Grid developers are developing tools for
  debugging and measuring the behavior of grid
  applications.

45
The present and the future

• Most grid systems include some job schedulers,
  but as grids span wider areas areas
• there will be a need for more meta-schedulers
  that can manage variously configured collections
  of clusters and smaller grids
• They will also extend their reach to implement
  better QoS, using
• reservations,
• redundancy,
• and history profiles of jobs and grid performance.

46
The present and the future

• Providing a reliable, well performing, and
  automatically recoverable virtual data sharing
  and storage.
• Projects are taking on this task, federating
  data, and achieving better performance,
  integration with scheduling, reliability, and
  capacity.
• Autonomic computing has the goal to make the
  administrators job easier by automating the
  various complicated tasks involved in managing a
  grid.
• These include identifying problems in real time
  and quickly initiating corrective actions before
  they seriously impair the grid.

47
A word of caution should be

• The grid is not a silver bullet that can take any
  application and run it a 1000 times faster
  without the need for buying any more machines or
  software.
• Not every application is suitable for running on
  a grid.
• Some kinds of applications simply cannot be
  parallelized.
• For others, it can take a large amount of work to
  modify them to achieve faster throughput.
• The configuration of a grid can affect the
  performance, reliability, and security of an
  organizations computing infrastructure.

48
References

1. V. Berstis, Fundamentals of Grid Computing IBM
   Redbooks paper