[cs615asa] HW#N

Xin Li xli76 at stevens.edu
Sun May 10 17:54:35 EDT 2015


Hi all,

I attended the meetup "Architecting for the Cloud: Avoiding Database
Disaster”. The speaker is Seth Proctor, he is currently CTO at NuoDB. NuoDB
is a cloud-based SQL database service provider.

He first talked about Cloud architecture:
----On-Demand Capacity: get resources when you need them, and shut them
down when you don't
----Simplicity: Use abstraction to simplify the programming and operation
experience
----Flexibility: Obtain resources easily
----Resiliency: Continuous availability and disaster recovery

Benefits:
----Greater capacity
----Higher availability
----Lower cost
----Global deployment
Challenge:
----Great failure
----Security

He explained the challenges for scaling database for the cloud, why the
traditional database didn’t work on cloud architecture.
----Inflexible
----Hard to scale out
----Hard to provide continuous availability
----Expensive to perform changes

Common options to resolve this problem:
----Apply replication: make copies of database
----Sharding: split database into smaller parts and store them in different
places.

Side effects of these options:
----Applications are not independent from database.
----Database become independent pieces that are hard to manage.
----Harder to interpret failure
----Complexity

He talked about the necessities of Global deployment: some governments
requires their citizens data center has to stay in their country, disaster
recovery, locality - reduce latency and cost.

He talked about 3 Distributed Database Designs:
----Shared-Disk, Bigger disk or disk array
----Synchronous Replication (expensive)
----Shared-Nothing/Sharded (unable to communicate with each other)

Then, he explained the architecting used by NuoDB - Durable Distributed
Cache:
----Cache: think database as a combination of service and data, use
in-memory database service to improve performance.
----Distributed: Database runs independently and able to coordinate with
each other to promotes resiliency and continuous availability.
----Durable: Disaster recovery and data continuous availability.

Cloud requirements:
----Scale on demand
----Provide continuous availability
----Simplified application and operation using abstraction
----Hybrid system, different locations, different environments
----Operate as a service
----Global operation

Video Link for meetup: https://www.youtube.com/watch?v=FDxBl-15iyE
Some Pictures of the meetup: http://www.cs.stevens.edu/~xli76/CS615/HWN.html

Best,
Xin
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.stevens.edu/pipermail/cs615asa/attachments/20150510/07ab8654/attachment.html>


More information about the cs615asa mailing list