Globus Monitoring and Discovery Service MDS

1
Globus Monitoring and Discovery Service (MDS)

• Jennifer Schopf
• Argonne National Laboratory
• June 14, 2002

2
MDS Monitoring and Discovery Service

• Globus Information Service
• Used by iVDGL, GriPhyN, PPDG, EDG, NMI, Grads,
  NorduGrid, etc.
• Requirements and characteristics
• Uniform, flexible access to information
• Scalable, efficient access to dynamic data
• Access to multiple information sources
• Decentralized maintenance
• Secure information provision
• Main contact John McGee (mcgee_at_isi.edu)

3
MDS Architecture

• Resources run a standard information service
  (GRIS) that speaks LDAP and provides information
  about the resource
• GIIS provides a caching service
• Resources register with GIIS
• GIIS pulls information when requested by a client
  (when out of date)
• GIIS provides the collective-level
  indexing/searching function

Client 1
Client 2
GRIS register with GIIS GIIS requests info from
GRIS services
Client 1 requests infodirectly from resources.
Client 2 uses GIIS for searching collective
information.
GIIS Cache contains info from A and B
4
Protocols

• MDS protocols based on LDAP
• Dynamic Registration via Reg. Protocol (GRRP)
• soft-state protocol
• Resource Inquiry via Info. Protocol (GRIP)
• Co-located with resource on network
• Resource Discovery (via GRIP or other)
• Using GRIP allows resource/directory hierarchy
• Also well defined interfaces to add new sensor
  data

5
Soft-state Registration

• Periodic notification
• Service/resource is available
• Expected-frequency metadata
• Automatic directory construction
• Add new resources to directory
• Invite resources to join new directory
• Self-cleaning
• Reduce occurrence of dead references

6
MDS-2 Implementation

• Grid Information Service (GRIS)
• Provides resource description
• Modular content gateway
• Grid Index Information Service (GIIS)
• Provides aggregate directory
• Hierarchical groups of resources
• NOTE EDG Ftree
• flexible backend for MDS by EDG WP3
• currently in EDG testbed installation of MDS

7
MDS-2.1 GRIS Providers

• grid-info-cpu reports CPU/load info
• grid-info-fs reports filesystem info
• grid-info-mem reports RAM/VM info
• grid-info-net reports NIC/net info
• grid-info-os reports OS info
• grid-info-platform reports arch. info
• grid-info-merged merges all host info
• Extensible for other sources, e.g. GRAM

8
Whats next for MDS (core)

• MDS 2.2 expected in July 2002
• Will harden the 2.1 code
• Additional performance improvements
• Address some of the packaging difficulties that
  occurred with 2.1
• Updated schemas (GLUE-schema project)
• Notification is currently not included in MDS 2.2
  because it was not required as part of the use
  cases gathered by the joint monitoring group
• The following release will likely be the
  OGSA-ified version, currently under discussion,
  and will include notification

9
2.1 GRIS Provider Times
10
GRIS Response Issues

• MDS 2.0 and 2.1 are lazy
• Probes are not issued unless queried
• Some system probes are slow
• Best probe may take several seconds
• How to avoid stale data?
• Clients set time-out per query
• GRIS/GIIS define time-out per source
• Fresh data found before timeout is returned
• Cache fill continues after client time-out

11
What next for MDS Monitoring

• Leading the effort to have common schemas between
  EDG and US projects
• Defining additional information providers in
  cooperation with sensor developers
• Eg. Using Ganglia for cluster-level monitoring

12
GridFTP Monitoring

• We added basic logging to GridFTP
• source address, file name, file size, number of
  parallel streams, TCP buffer size, start and end
  timestamps, total time consumed, aggregate
  bandwidth achieved, read/write
• Theres also an information provider available
• www-unix.mcs.anl.gov/vazhkuda/GridFTP-Information
  -Provider/
• And a paper talking about predictions
• www.globus.org/research/papers/Prediction-Paper-24
  9.pdf

13
Things to think about when comparing systems

• What is the main use case your system addresses?
• Resource selection
• What are the base set of sensors given with a
  system?
• (see several slides ago)
• How does that set get extended?
• Well defined process for wriring an information
  provider to act as a front end to any sensor and
  translate the data into ldap see
  www.globus.org/mds/creating_new_providers.pdf
• What are you doing for discovery/registry?
• We use an embeded discovery/registry

14
Comparing 2

• What schema are you using (do you interact with)?
• Globus defined its own schema for MDS 2.1
  release, will be using GLUE-schema as well in MDS
  2.2
• Is this system meant to monitor a machine, a
  cluster, or send data between sites, or some
  combination of the above?
• This system interfaces to machine, cluster or
  site data, and communicates that information
  between sites. It is not meant to replace a site
  monitoring system. It is meant to be used to have
  a common infrastructure to allow interoperablity
  between site monitoring systems

15
Comparing 3

• What kind of testing has been done in terms of
  scalability
• This is in progress- but the software is
  currently being used by iVDGL, GriPhyN, PPDG,
  EDG, NMI, NPACI, NCSA, Grads, NorduGrid, etc.

16
Glue Schema Status

• Jennifer Schopf
• Argonne National Lab
• May 23,2002

17
GLUE-Schema Effort

• Part of HICB/JTB GLUE framework
• To address need to common schemas between
  projects
• framework independent
• something to translate into, not a requirement
  within fabric layer
• maillist glue-schema_at_hicb.org
• www.hicb.org/glue/glue-schema/schema.html

18
Problems with Schemas

• Naming MetaData implies a structure, e.g.
  relational, hierarchical, etc
• Attribute/Measurement units
• Attribute/Measurement naming
• Object/Entity naming
• Representation of a schema (LDAP, SQL, XML, CIM,
  etc)
• A lot of schemas out there
• information, monitoring, scheduling, accounting,
  security, and more!

19
Goal

• Define a minimum common schema requirement for
  interoperability
• Compute Elements, Network Elements, Storage
  Elements
• EDG, DataTag, PPDG, GriPhyN, iVDGL, etc.!
• First step of this achieved
• CE schema currently being implemented by both
  Globus and EDG information providers
• SE schema currently under discussion on mailing
  list

20
Spreadsheet version 12

• Listing EDG, Globus and new schemas for CE
• First draft March 19
• 12 calls, more than 500 emails, and MANY hours
  of work later here we are!
• Draft 12 sent out May 22 pm

21
Document v. 6

• Layered structure CE, cluster, (homogeneous)
  subcluster, node
• Attributes collected into object classes
• UML class diagram for generic representation
• Currently being mapped into ldap/sql/xml

22
GLUE-Schema Structure
23
Status

• Implementation of information providers beginning
• MDS information providers with new schemas part
  of the MDS 2.2 release, mid July
• EDG information providers need updating as well
• Test between EU and US planned for July

24
After CE we have SE and NE

• Discussion of SEs has started
• Experts have been asked to join in
• Arie Shoshani, John Gordon
• Glue Storage Model beginning to be discussed to
  understand general constraints
• NE will be addressed after this

25
Additional Information

• MDS information
• www.globus.org/mds/
• GLUE-schema
• www.hicb.org/glue/glue-schema/schema.html