Network Performance Monitoring - PowerPoint PPT Presentation

1 / 8
About This Presentation
Title:

Network Performance Monitoring

Description:

13th GridPP Collaboration Meeting, Durham, 4 6 July 2005. Publication Schema ... to concatenate data from several backbone networks to produce a picture of the ... – PowerPoint PPT presentation

Number of Views:16
Avg rating:3.0/5.0
Slides: 9
Provided by: RobinT7
Category:

less

Transcript and Presenter's Notes

Title: Network Performance Monitoring


1
Network Performance Monitoring
  • Robin Tasker
  • CCLRC, Daresbury Laboratory
  • 5 July 2005

13th GridPP Collaboration Meeting, Durham, 4 6
July 2005
2
Publication Schema
  • Request / Response Approach
  • Interactive systems clients will be able to
    request
  • historic data,
  • future or on-demand tests
  • predictions (i.e. NWS style).
  • All request and result messages are
  • formatted using standardised schemas
  • developed within the GGF NMWG.
  • This allows heterogeneous monitoring
  • systems to interact providing that they
  • use the same schemas.

3
The ad hoc Approach
If a Client, whatever that may be, wishes to
receive detailed information concerning a network
path or to concatenate data from several
backbone networks to produce a picture of the
full backbone path. Then the Client has to make
several requests, and if necessary, perform its
own data aggregation.
4
The Mediator Approach
EGEE JRA4 is responsible for Development of
Network Services. Within JRA4s Network
Performance Monitoring (NPM) activity, the
Mediator software has been developed. This
will greatly simplify this process by unifying
access to network performance measurement data.
5
The Way We Were...
6
And How We Will Be 1
  • Work in Progress
  • An infrastructure in which each site
  • (MP) will store its test data in a central
    database.
  • 2. Web Services and human (web) access to the
    data will also be via services running centrally.
  • 3. Storing and providing access to the data from
    a central location reduces complexity of the
    individual monitoring nodes
  • 4. Critically, we can also move to a relational
    database model, speeding up access to the data
    and allowing considerably more advanced queries
    to be made

7
And How We Will Be 2
  • Use of standard package management techniques,
  • the monitoring node software available as a set
    of RPMs
  • a Yum repository created to distribute initial
    installations and subsequent updates to the
    infrastructure.
  • To avoid creating a single point of failure, both
    the data and access software can be mirrored at
    another well known location.
  • Plan to locate an MP at the Tier 1 and each of
    the Tier 2 institutes and in addition, through
    JISC support and working with UKERNA, to locate
    MPs within the national network infrastructure

8
Discuss
Write a Comment
User Comments (0)
About PowerShow.com