gc3: grid computing competence center gc3: grid computing competence center grid computing riccardo

Download GC3: Grid Computing Competence Center GC3: Grid Computing Competence Center Grid Computing Riccardo

Post on 13-May-2020

5 views

Category:

Documents

1 download

Embed Size (px)

TRANSCRIPT

  • GC3: Grid Computing Competence Center

    Grid Computing Riccardo Murri, Sergio Maffioletti GC3: Grid Computing Competence Center, University of Zurich

    Oct. 31, 2012

  • Grid Computing (the vision)

    “A computational grid is a hardware and software infrastructure that provides dependable, consistent, pervasive, and inexpensive access to high-end computational capabilities.”

    Reference: Foster, I., Kesselman, C., “Computational Grids”, in: “The Grid:

    a Blueprint for a New Computing Infrastructure”, Morgan-Kauffman,

    1999.

    LSCI2012 Grid Computing Oct. 31, 2012

  • Grid Computing (the vision)

    “A computational grid is a hardware and software infrastructure that provides dependable, consistent, pervasive, and inexpensive access to high-end computational capabilities.”

    Predictable and sustained level of performance.

    LSCI2012 Grid Computing Oct. 31, 2012

  • Grid Computing (the vision)

    “A computational grid is a hardware and software infrastructure that provides dependable, consistent, pervasive, and inexpensive access to high-end computational capabilities.”

    Standardized interfaces/APIs

    LSCI2012 Grid Computing Oct. 31, 2012

  • Grid Computing (the vision)

    “A computational grid is a hardware and software infrastructure that provides dependable, consistent, pervasive, and inexpensive access to high-end computational capabilities.”

    Always available in a large variety of environments.

    LSCI2012 Grid Computing Oct. 31, 2012

  • Grid Computing (the vision)

    “A computational grid is a hardware and software infrastructure that provides dependable, consistent, pervasive, and inexpensive access to high-end computational capabilities.”

    Affordable and convenient (relative to the delivered performance).

    LSCI2012 Grid Computing Oct. 31, 2012

  • HPC vs HTC

    High-Performance Computing Minimize turnaround time of computational applications.

    High-Throughput Computing Maximize amount of “work” done in a given time.

    LSCI2012 Grid Computing Oct. 31, 2012

  • HPC vs HTC

    Given a fixed set of resources, the two objectives are equivalent.

    Fast wide-area networking and cheaper computing hardware (and now even cheaper IaaS provisioning) enable aggregation of geographically distributed compute resources as HTC facilities.

    LSCI2012 Grid Computing Oct. 31, 2012

  • Grid Computing (the reality)

    An aggregation of computational clusters for execution of a large number of batch jobs.

    LSCI2012 Grid Computing Oct. 31, 2012

  • HTC system stakeholders

    owners Set the policy for the usage of a resource

    sysadmins Take technical and operational decisions

    developers (application writers) Interact with HTC system via its APIs and adapt applications to its conceptual model.

    users (customers) Need applications to work; HTC system must be flexible enough to adapt to their requirements.

    LSCI2012 Grid Computing Oct. 31, 2012

  • Example: the UZH “Schrödinger” cluster

    Informatik Dienste UZH

    { owner

    admin

    Research groups

    { developers

    users

    LSCI2012 Grid Computing Oct. 31, 2012

  • Example: SMSCG

    Collaborative project to share HPC clusters

    among higher-education institutions.

    http://www.smscg.ch/

    LSCI2012 Grid Computing Oct. 31, 2012

    http://www.smscg.ch/

  • Example: SMSCG

    owners Participating institutions

    sysadmins Personnel of the participating institutions

    developers Support groups (e.g., GC3)

    users Research groups

    LSCI2012 Grid Computing Oct. 31, 2012

  • Resource Management layers

    user Run tasks on the distributed infrastructure; application-level scheduling and resource selection.

    grid

    – Provide uniform/abstract view of computational resources and authentication/authorization credentials.

    – Allocate resources to tasks.

    fabric Actual execution of tasks and storage of data, according to owner policies.

    LSCI2012 Grid Computing Oct. 31, 2012

  • Scheduling on a cluster

    compute node N

    local 1Gb/s ethernet network

    internet

    batch system server

    compute node 2compute node 1

    ssh username@server

    �� �� �� ���� �� �� ��

    All job requests sent to a central server.

    The server decides which job runs where and when.

    LSCI2012 Grid Computing Oct. 31, 2012

  • where: resource allocation model

    Computing resources are defined by a structured set of attributes (key=value pairs).

    SGE’s default configuration defines 53 such attributes: number of available cores/CPUs; total size of RAM/swap; current load average; etc.

    A node is eligible for running a job iff the node attributes are compatible with the job resource requirements.

    (Other batch systems are similar.)

    LSCI2012 Grid Computing Oct. 31, 2012

  • when: scheduling policy

    There are usually more jobs than the system can handle concurrently. (Even more so, in high-throughput computing cases we are interested in.)

    So, job requests must be prioritized.

    Prioritization of requests is a matter of the local scheduling policy.

    (And this differs greatly among batch systems and among sites.)

    LSCI2012 Grid Computing Oct. 31, 2012

  • (Hidden) assumptions

    1. The scheduling server has complete knowledge of the nodes

    Local networks have low latency (RTT average 0.3 ms on a 1GB/s ethernet) and the status information is a small packet.

    2. The server has complete control over the nodes

    So a compute node will immediately execute a job when told by the server.

    LSCI2012 Grid Computing Oct. 31, 2012

  • How does this extend to Grid computing?

    By definition of a Grid. . .

    1. It’s geographically distributed – High-latency links (hence: resource status may be

    not up-to-date)

    – Network is easily partitioned or nodes disconnected (hence: resources have a dynamic nature; they may come and go)

    2. Resources come from multiple control domains – Prioritization is a matter of local policy!

    – AuthZ and other issues may prevent execution at all.

    LSCI2012 Grid Computing Oct. 31, 2012

  • The Globus/ARC model

    compute node N

    local 1Gb/s ethernet network

    batch system server

    compute node 2compute node 1

    compute node N

    local 1Gb/s ethernet network

    batch system server

    compute node 2compute node 1

    compute node N

    local 1Gb/s ethernet network

    batch system server

    compute node 2compute node 1

    internet

    arcsub/arcstat/arcget

    �� �� �� ���� �� �� ��

    An infrastructure is a set of independent clusters.

    The client host selects one cluster and submits a job there. Then periodically polls for status information.

    LSCI2012 Grid Computing Oct. 31, 2012

  • Issues in the Globus/ARC approach?

    1. How to select a “good” execution site?

    2. How to gather the required information from the sites?

    3. Based on the same information, two clients can arrive on the same scheduling information, hence they can flood a site with jobs.

    4. Actual job start times are unpredictable, as scheduling is ultimately a local decision.

    5. Client polling increases the load linearly with the number of jobs.

    LSCI2012 Grid Computing Oct. 31, 2012

  • The MDS InfoSystem, I

    compute node N

    local 1Gb/s ethernet network

    batch system server

    compute node 2

    GRIS

    compute node N

    local 1Gb/s ethernet network

    batch system server

    compute node 2

    GRIS

    compute node N

    local 1Gb/s ethernet network

    batch system server

    compute node 2

    GRIS

    internet

    arcsub/arcstat/arcget

    �� �� �� ��

    �� �� �� ��

    �� �� �� ��

    �� �� �� ��

    �� �� �� ��

    �� �� �� ��

    �� �� �� ��

    �� �� �� ��

    �� �� �� ��

    �� �� �� ��

    �� �� �� ��

    �� �� �� ��

    ����� ����� ����� ����� ����� ����� ����� �����

    ����� ����� ����� ����� ����� ����� ����� �����

    ����� ����� ����� ����� �����

    ����� ����� ����� ����� �����

    ����� ����� ����� �����

    � � � � � � � � � �

    � � � � � � � � � �

    � � � � � � � � �

    � � � � � � � � �

    � � � � � � � � �

    � � � � � � � � �

    � � � � � � � � � �

    � � � � � � � � � �