Transcript G19 - Spatial Database Group

Cloud
Databases
Matt Gregg
Bob Guidinger
Cloud 101
• What do we mean by Cloud
Databases?
• Why do we have them?
o Alternative to IT infrastructure investment
(pay-as-you-go)
o Analytical Usage: Data warehousing, data mining (readintensive)
o Huge, fast storage, grid computing, virtualization, N-tier
architecture, robust networks.
o Elasticity, scalability, high availability, price-per-usage and
multi-tenancy (fault tolerance)
o Access anywhere, at any time, from any device
Types
• Cloud Storage
o Files stored in the cloud
o iCloud, Dropbox, etc.
• Data as a Service (DaaS)
o Data stored in the cloud
o Backups
• Database as a Service (DBaaS)
o Data stored in the cloud
o Full database management functionality
o Common names
• Amazon SimpleDB, Amazon RDS, Google BigTable, Yahoo
Sherpa and Microsoft SQL Azure Database
Storage Architectures
• Shared-nothing
o Splits the data into independent sets stored physically on different
servers
o Easily scalable
o Difficult to maintain in cloud with data partitioning (shipping
latency)
o Ex. Oracle, Hadoop, Amazon’s Simple DB
• Shared-disk
o
o
o
o
Data stored on a SAN or NAS
Fewer, low-cost servers
Easy to virtualize
Access to all data
Transaction Processing applications. Oracle RAC, IBM
DB2 pureScale, Sybase etc. support this architecture [11].
Distributed
Scalability
ACID
OLTP
Analytical
Maintenance
Cost
Useful for
Cloud
SharedNothing
SharedDisk
Partitioning
Architecture
Table 1: Comparison of shared-nothing and shared disk storage
architectures
Y
Y
Y
N
N
Y
High
Y
N
Y
Y
Y
Y
Y
Low
Y
Note: N-No, Y- Yes
3. A Comparative Study of Relational
Challenges for Cloud
Databases
• ACID vs. BASE (Basically Available, Soft state,
Eventually consistent)
• Transaction processing
o Data Consistency/Integrity
o Database Security/Privacy
• Analytical Processing Challenges
o
o
o
o
Developing Scalability
Querying a distributed database
Vender portability
Heterogeneous
Cloud Database Providers
• Note: These are managed database
services.
• SQL Services
o Amazon Rel. Database Service, Clustrix, EnterpriseDB
PostgreSQL, Google Cloud, HP Cloud Rel. DB, IBM
SmartCloud, Microsoft SQL Azure, Oracle DB Cloud,
Xeround
• NoSQL Services
o Amazon DynamoDB, Amazon ElastiCache, Cloudant,
Database.com, Microsoft Azure Table Storage, MongoHQ
Summary
• Changing use of Databases
• Benefits
o Elastic, scalable, cheap, fault tolerant, access
anywhere
• Cloud Storage, DaaS, DBaaS
• Shared Nothing vs. Shared Disk
• BASE
References
1. Arora, Indu, and Anu Gupta. "Cloud Databases: A
Paradigm Shift in Databases.” International Journal of
Computer Science Issues. 9.4 (2012): 77-83. Web. 22
Apr. 2013. <http://ijcsi.org/papers/IJCSI-9-4-3-7783.pdf>.
2. Harris, Derrick. "Cloud databases 101: Who builds ‘em
and what they do." GIGOM. N.p., 20 JULY 2012. Web. 22
Apr 2013. <http://gigaom.com/2012/07/20/clouddatabases-101-who-builds-em-and-what-they-do/>.
3. Bridgwater, Adrian. "Cloud databases: are lazy
developers cutting corners?." Cloud Pro, 13 MAR 2013.
Web. 22 Apr. 2013. <http://www.cloudpro.co.uk/adrianbridgwater/5378/cloud-databases-are-lazy-developerscutting-corners>.