Title of Talk:   A Method to Protect Privacy of RDBMS Data on a Data Grid



Abstract:

The massive investments made in the area of telecommunications has left us

in a glut of bandwidth.  Grid computing is a model well suited to leverage

this glut.  The  shared network storage model may be viewed as a data grid

in evolution.  Database Management Systems (DBMS) have been traditionally

implemented assuming a privately attached disks.  In a data grid, or

networked storage, a DBMS may use locally attached, locally administered

private storage or may take advantage of the  networked storage, typically

administered by another party.  The privacy problem arises if this party

may not be trusted, or if this party, even if trustworthy, does not have

sufficient security services in place.



Encryption is a standard means of providing privacy.  Encrypting data

before storing it is a good approach.  However, most data applications

apply logic to data before delivering results.  In a client server model,

logic may be applied on the server side and/or the client side.  If we are

to store encrypted data on the server, assume that only the user on the

client is privy to the decryption key, the model implies that, general

logic can not be applied at the server.  Not only does this contribute to

under utilization of the server but it has negative implications on

contention of shared data due to the increased granularity of locking and

increased latencies.



However, it was discovered that if we take a subset of logic, say the logic

expressible by SQL, it is possible to apply a significant portion of this

logic on the server, with the rest of the logic being applied on the client

after decryption.  This talk will discuss how this may be done.



Bio: IBM Distinguished Engineer, Bala Iyer, manager of the Advanced

Technology Development group at the IBM's Silicon Valley Lab, earned his

Masters and PhD from Rice University, winning the Best Thesis Award at the

university.  In the past has worked as a Member of Technical Staff at Bell

Labs in Murray Hill, NJ and as a Research Staff Member at IBM TJ Watson

Research Center, NY. He has devoted the last 15 years to R&D for IBM Data

Management products.  Bala is the co-architect of IBM z-Series server's

hardware compression facility, and compression software support in IBM's

DB2, IMS and VSAM software products.  Bala has contributed in numerous

other areas  including the data grid, data security and privacy,

application service provisioning, storage for Life Science, e-business, ERP

application performance, data mining, multidimensional analysis, SQL

optimization, database performance, high performance sorting and

synchronization algorithms for scalable database clustering, analytical

performance modeling, high speed network switching, and computer network

architecture.  Recently, he architected and built a special computing

system for the Protein Data Bank to store and process protein structure

data.  Bala has published extensively and has been granted over 50 US

patents.