An Overview of the Federated MCAT design - PowerPoint PPT Presentation

1 / 25
About This Presentation
Title:

An Overview of the Federated MCAT design

Description:

Create foo in mcat Zone 'z1', in resource 's1' of 'z1' ... SRB server queries mcat 'z1' for metadata for file /z1/x/y/z/foo ... collectionName - use z option ... – PowerPoint PPT presentation

Number of Views:22
Avg rating:3.0/5.0
Slides: 26
Provided by: sdsc
Category:

less

Transcript and Presenter's Notes

Title: An Overview of the Federated MCAT design


1
  • An Overview of the Federated MCAT design
  • (SRB 3.0)
  • By Michael Wan
  • http//www.npaci.edu/DICE/SRB/

SDSC/UCSD/NPACI
2
Single MCAT architecture
  • Federated middleware system
  • Single MCAT server
  • Metadata catalog
  • Use traditional DBMS
  • Federated resource servers
  • Single client signOn on anyone of the servers
  • Access to all resources in the federation
  • Robust server-server operation

3
Single MCAT architecture
Peer-to-peer Brokering
Client Read
Parallel Data Access
Logical Name Or Attribute Condition
1
6
5/6
SRB server
SRB server
3
4
5
SRB agent
SRB agent
2
Server(s) Spawning
R1
MCAT
1.Logical-to-Physical mapping 2. Identification
of Replicas 3.Access Audit Control
R2
Data Access
4
Single MCAT Concepts and Features
  • A single global User Space
  • Single sign-on, access all resources
  • Single class of Admin User
  • Multiple authentication schemes
  • GSI, secure passwords, tickets
  • Robust access control

5
Single MCAT Concepts and Features (cont1)
  • Data and Collections
  • Logical Name space -
  • UNIX like directories (collections) and files
    (data)
  • Mapping of logical name to physical attributes -
    host address, physical path.
  • UNIX like API and utilities for making
    collections (mkdir) and data creation (creat)
  • Virtualisation of Resources
  • Mapping of a logical resource name to physical
    attributes Resource Location, Type
  • Client use a single logical name to specify a
    resource

6
Single MCAT Concepts and Features (cont2)
  • Performance enhancement
  • Client and Server-driven Parallel I/O strategies
    (3-4 times)
  • Interface with HPSSs mover protocol for parallel
    I/O
  • Parallel third party transfer for copy and
    replicate
  • One hop data transfer between client and data
    resource
  • Container physical grouping of small files for
    tape I/O
  • Bulk load and unload speed up uploading and
    downloading of small files (10-50 times)

7
Motivation for a Federated (multiple) MCAT system
  • MCAT performance network latency for a world
    wide network.
  • East/west coast latency for a simple query 1-2
    sec.
  • Local control
  • Security issue multiple admin users from
    different sites
  • organizations want to have more control on the
    data and resources that they are sharing
    worldwide.
  • Scalability of the MCAT.
  • No single point of failure

8
Design goal of 3.0
  • Very Basic features - incremental changes
  • Multiple MCATs
  • MCAT ZONE a new term
  • defines a federation of SRB resources controlled
    by a single MCAT
  • Each Zone has its own sys admin - local control
    of users and resources
  • Peer to peer zone architecture
  • Each Zone can operate entirely independently from
    other zone.
  • Data and Resource sharing across ZONES
  • Use storage resources in foreign zones
  • Access data stored in foreign zones
  • Copy data across zones

9
Peer to peer Federated MCAT
MCAT1
Server1.1
MCAT3
Server1.2
Server3.1
MCAT2
Server2.2
Server2.1
10
Peer to peer Federated MCAT implementation
  • Development tasks needed
  • 1) New metadata for Zones
  • new Zone table
  • add zone to user metadata
  • 2) Authentication across zones
  • 3) Resource and data access across zones
  • Version 3.0 - Unix server, Scommand client only

11
1) Zone metadata implementation (1)
  • System of MCAT zones
  • Each MCAT manages most metadata independently of
    other zones
  • No change for most metadata and metadata
    operations
  • New meatdata
  • Zone Info - new table defines all zones in the
    federation
  • zoneName logical name
  • Network address
  • Local/Foreign
  • Authentication info
  • Each MCAT maintains a complete zone table

12
1) Zone metadata implementation (2)
  • User Info - defines all users in the federation
  • Single global user name space
  • Each MCAT maintains a table of all users
  • Local/Foreign user a user local to only one
    zone
  • Sensitive info (password) only stored in users
    local zone

13
1) Zone metadata implementation (3)
  • New Admin tool for handling new metadata
  • Zone Authority page
  • obtain a unique zoneName
  • Web page available with 3.0 release
  • Create and modify Zone metadata
  • Java admin GUI , command line
  • User Admin tool
  • Modified to handle Zone related metadata
  • New tools to poll foreign MCATs for new users and
    add to local MCAT.

14
2) Cross Zone Authentication (1)
  • Support 2 authentication schemes
  • Encrypted Password
  • challenge-response scheme
  • no password over the net
  • GSI
  • Phased out plain text password and SEA

15
2) Cross Zone Authentication (2)
  • Support a robust set of server-server operations
  • servers running as privileged SRB users perform
    operations on behalf of client users.
  • Want to limit privilege of admin user from
    foreign zone
  • an admin user can only requests a foreign zone to
    perform operations behalf of client users from
    the SAME zone
  • sensitive info - password of local user stored
    only in local MCAT
  • security compromise in one zone cannot spread to
    other zone

16
2) Cross Zone Authentication (3)
  • Implication of security measure
  • Lost a little transparency.
  • Users must first connect to a server in its own
    zone first for across zone operation
  • Only your own admin user can go across zone for
    you
  • Very minor effect on data operation

17
3) Resource and data access across zones (1)
  • Make easy by robust server-server support
  • Most existing operations are supported across
    zones
  • put, get, copy, bload, bunload, register,
    container operations, etc
  • replication across zone not supported (use Scp
    for now)
  • Typical operation open a collectionName /x/y/z
    for read.
  • Data server queries MCAT for location, file type,
    etc
  • But which MCAT to query ?
  • The first entry in the collectionName specifies
    the zoneName where the metadata is stored
  • for /z1/x/y/z z1 is the zoneName
  • Similar to mount point in FS
  • Most data handling codes unchanged. New codes to
    determine which MCAT to go to.

18
3) Resource and data access across zones (2)
  • Most work involves adding logic to determine
    which MCAT to go to.
  • Sput S s1 foo /z1/x/y/z
  • (Scd /z1/x/y/z Sput S s1 foo .)
  • Upload the local file foo to SRB.
  • Create foo in mcat Zone z1, in resource s1 of
    z1
  • SRB server query mcat z1 for metadata for
    resource s1 (network address, file type)
  • Upload foo and put it in resource s1
  • Register the file /z1/x/y/z/foo with mcat z1

19
3) Resource and data access across zones (3)
  • Sget /z1/x/y/z/foo .
  • Download the SRB file foo to local FS
  • SRB server queries mcat z1 for metadata for
    file /z1/x/y/z/foo
  • Discover file is in resource s1.
  • Request resource s1 server to download the
    file.

20
3) Resource and data access across zones (4)
  • Scp S s2 /z1/x/y/z/foo /z2/a/b/c
  • Copy the SRB file foo managed by mcat z1 to the
    resource s2 of mcat z2
  • Query z1 for file foo, found file in resource
    s1. Query z2 for resource s2
  • Copy the file foo stored in resource s1 to
    sources2.
  • Register the file /z2/a/b/c with mcat z2

21
3) Resource and data access across zones (5)
  • Some Scommand does not involve collectionName -
    use z option
  • -z zoneName option - explicitly specify a
    zoneName
  • Use cwd if -z option not used
  • SgetR -z z1
  • Smkcont -z z1 -S s1 cont1
  • If cwd/z2/x/y/z, Slscont

22
3) Resource and data access across zones (6)
  • Registration of data files in more than one Zone
  • File created/registered in zone z1
  • /z1/u.d/x/y in resource s1 in zone z1
  • Register also in zone z2
  • /z2/u.d/x/y in resource s1 in zone z2
  • s1 should be known to z2
  • Note that collectionName changes.
  • Some system-metadata carried over when doing
    inter-zone registration
  • Copy user-defined metadata across the zones (if
    needed)
  • Lazy synchronization (user-controlled)
  • In 3.0
  • Zone z1 and z2 are unaware of each others copies
  • In later versions - awareness improved
  • Metadata synchronization
  • Deletion notification
  • User-controllable delay in synchronization

23
Zone A
Zone B
/A/u3
/A/u1
/B/u1
/B/u3
/A/u4
/A/u2
/B/u2
/B/u4
/B
/A
/A/u1/f1? R1P1
/B/u3/f2? R2P2
/A/u3/f2? R2P2
/B/u1/f4? R1P4
User Synchronization
/A/u2/f1? R1P3
User Synchronization
Resource Synchronization
Zone Synchronization
DataSynchronization
24
3.0 Status
  • Slightly delayed
  • All core development done
  • Zone metadata
  • cross zone authentication
  • cross zone resource and data access
  • Items not yet complete
  • User management tool - another week
  • Documentation
  • testing
  • Release date - Later part of September

25
Future plans
  • 3.1 - The rest of SRB software - window, inQ,
    mySRB, Jargon and Matrix
  • 3.2
  • use of proxy certificate for cross zone
    authentication
  • Kerstin Kleese group at DL
  • Cross zone replication
  • Synchronization of copies
  • versioning
  • locking across zone ?
  • Cross zone soft link
Write a Comment
User Comments (0)
About PowerShow.com