More Related Content
Similar to SQL and NoSQL Databases in the Cloud Conference
Similar to SQL and NoSQL Databases in the Cloud Conference (20)
More from Romeo Kienzler (20)
SQL and NoSQL Databases in the Cloud Conference
- 1. The Data Warehouse Institute DW Conference 2014
Zurich – 17/18.11.14
bit.ly/go4bluemix
© 2013 IBM Corporation
SQL and NoSQL Databases in the cloud
Twitter: @RomeoKienzler
Romeo Kienzler – IBM Innovation Center Zurich
- 2. BlueMix is a Platform as a Service Cloud,
based on Cloud Foundry, employing Enterprise
grade services enriched with IBM Software and
hosted at SOFTLAYER
bit.ly/go4bluemix
2 © 2013 IBM Corporation
- 5. Advantages of Database-as-a-Service (DBaaS)
Design
Admin
Software
Hardware
Design
Admin
Software
Design
Guesswork, Overhead,
Risk
5
✓ Guaranteed
✓ No Guesswork
✓ Smallest Risk
✓ Greatest Productivity
Do-It-Yourself Hosted DBaaS
© 2014 IBM Corporation
- 6. Cloudant is Used by Big Web & Big Mobile Leaders
• 17,000 users
• Billions of
daily transactions
• Millions of DBs
• Across 34 DCs on
5 Cloud providers
eCommerce
Online Education
Gaming
Financial Services
Networking
Other Industries
© 2014 2013 IBM Corporation
- 7. Why cloud?
● Elasticity of
● Storage
● → Increase Ratio „disks per system“
● Processing power
● → Increase Ratio „cpu per disk“
● Cache
● → Increase Ratio „memory per storage unit“
● Automatic Backups
● Automatic Disaster Recovery
● Automatic Maintenance
7 © 2013 IBM Corporation
- 8. Implications?
● Data partitioning
● Complete data corpus resides on multiple
● Systems
● Disks
● Query / Transaction touches multiple
● Systems
● Disks
● Disk and System failures during runtime
● „Hot“ System and Disk removement / addition
8 © 2013 IBM Corporation
- 9. ACID vs CAP vs BASE
● ACID
● Atomicity (all or nothing)
● Consistency (never leave system in invalid state)
● Isolation (cope with nondeterminism)
● Durability (never loose data after commit)
● CAP Theorem
● Consistency (basically what ACID describes)
● Availability (always respond to every request)
● Partition Tolerance (tolerant to node failure)
● BASE
● Basically Available (but response negative/wait)
● Soft State (not always 100% consistent state)
● Eventually Consistent (on stop, sooner or later)
9 © 2013 IBM Corporation
- 10. Isolation vs. Consistency
● ACID
● Read Uncommited
● Read Commited
● Repeatable Read
● Serializable
● BASE
● Monotonic read
Once read, never read older version
● Monotonic write
Write order in all replicas is preserved
● Read your own writes
Once written, never read older version
10 © 2013 IBM Corporation
- 11. Live Demos
● You choose now using the Voting – APP :)
bit.ly/go4bluemix
11 © 2013 IBM Corporation