Structured Data Archiving in the Cloud

1
Structured Data Archiving Key Issues 9th July
2009 CloudCamp, London Mark
Cusack Principal Architect RainStor
2
Ideal Use Case Application Retirement
Cloud Bursting e.g reporting peak processing
Usage
Mission Critical Production System - On-premise
Dev Test
Retirement
Implementation
End-of-Life
Life Cycle
3
Data Volume Issues

• How to transfer in terabytes of data?
• Structured data can be massively compressed
  before uploading to a storage cloud, e.g. Amazon
  S3

4
Data Security Issues

• How to maintain data privacy and integrity?
• Theoretical solution homomorphic encryption!?
• Reality encrypt network pathways and data
  rest-points
• Blind cloud storage
• Key should be generated by the application owner
• Data should be encrypted on-premise, prior to
  transfer to the cloud
• Tamper-proofing and auditing
• Keep digests of database files off-cloud
• Practical in the case of application retirement
  (read-only)

5
Data Availability Issues

• How to ensure that the data is always available?
• Make multiple copies of the data
• Data compression makes this cost-effective
• Employ emerging cloud interoperability standards
• De facto standards Eucalyptus, Amazon Web
  Services
• SNIA eXtensible Access Method (XAM) as an
  alternative cloud storage interoperability
  standard

6
Data Query Issues

• How to query the data without compromising
  performance, security and accessibility?
• Use a compute cloud to query the data
• E.g. run retired application reports on EC2
  against S3 data
• Query directly against compressed data
• Problem becomes CPU-bound rather than IO-bound
• Pass encryption key to the cloud on a per-query
  basis
• Provide an ODBC/JDBC interface to compute
  instance

7
More Information and Contact Details

• www.rainstor.com
• twitter.com/rainstor
• twitter.com/markcusack
• mark.cusack_at_clearpace.com