Send the link below via email or IMCopy
Present to your audienceStart remote presentation
- Invited audience members will follow you as you navigate and present
- People invited to a presentation do not need a Prezi account
- This link expires 10 minutes after you close the presentation
- A maximum of 30 users can follow your presentation
- Learn more about this feature in our knowledge base article
MSc Thesis Presentation
Transcript of MSc Thesis Presentation
(visioncloud.eu) is a FP-7 funded EU research project with a focus is on innovative cloud storage."
Content Centric Service (CCS)
, which implements one of the innovations of Vision Cloud, is
a cloud access service
written in Java and an innovative next generation cloud service enabling user to describe the content through
and then access to the content by using this metadata
In a content centric approach, the user will query the data by using its
metadata, instead of by using its path in the hierarchical tree
Data object = Data + Metadata
Option 2 (better): Move to public cloud!
- Windows Azure
content centric service
Prof. Michael Gerndt
Prof. Carsten Trinitis
Enver Bahar ( )
25 + min
Intro - Hybrid what?
Content Centric Service
Background - Hybrid Cloud
A hybrid cloud environment is a composition of
two or more
of multiple deployment models
Background - Hybrid Cloud 2
Background - CCS 1
offer security at network level, and
offer scalability and cost efficiency.
is a combination of both of them
The approch of keeping some data
and using the
of public cloud is much more practical for many business cases i.e.
Intro - Idea
A LOT OF DATA
Company has an in-house storage, a small cloud. As company gets bigger the need for data as well.
Option 1: Buy new hardware!
- Expensive and fixed costs
- Hard to maintain
- Administration costs
Try to use public cloud and private cloud together
Have your critical data in-house, and non-critical data on public cloud vendors!!!
You are running a company
growing really fast
Background - CCS 2
YouTube metadata fields
why not for all kinds
Background - CCS 3
Youtube forces users to upload some metadata along with the data so that user can query by metadata
CCS aims to build that to all kinds of data (text, image, video ...)
CCS enables extensive queries for all kinds of data
CCS enables user defined metadata schemas
CCS enables usage of
kind of storage thus preventing vendor lock-in
storage access layer
Background - CCS 4
Background - CCS 5
Architecture - Sharding
How to distribute data between the clouds?
is a type of database partitioning that separates very large databases the into smaller, faster, more easily managed parts
Different sharding mechanisms:
Key based partitioning
Range based partitioning
Hash based partitioning
Directory based partitioning
Architecture - Implementation
How to make one cloud to be aware of the other cloud?
: A data structure that contains metadata about remote data rather than the actual data (
Holds metadata about clouds such as access credentials
A management layer above the cloud systems and stores information(FederationInfo) about other clouds
Client side app enables a unified view among the clouds
Defines how data is sharded and queried
"eu.visioncloud.federation.local_cloud url" : "vision-tb-1.cloudapp.net",
"eu.visioncloud.federation.target_cloud url" : "vision-tb-2.cloudapp.net",
"eu.visioncloud.federation.target_cloud port" : "8080",
"eu.visioncloud.federation.target_cloud ccs uri" : "/CCS/",
"eu.visioncloud.federation.local_cloud port" : "8080",
"eu.visioncloud.federation.local_cloud ccs uri" : "/CCS/",
"eu.visioncloud.federation.target_container name" : "vision2",
"eu.visioncloud.federation.local_container name" : "vision1",
"eu.visioncloud.federation.type" : "sharding",
"eu.visioncloud.federation.private cloud" : "vision1",
"eu.visioncloud.federation.public cloud" : "vision2"
"confidential" : "true"
"confidential" : "false"
Results - Single Cloud 1
Youtube Deutsche Welle channel
of metadata belonging to
Each video containing
Made several number of queries at the same time
queries respectively at once to show whether single response time will degrade with increasing number of requests
A pool of 10 threads
Single cloud tests
Results - Single Cloud 2
Same dataset, divided into
so that one cloud having 23206 the other 23207
Same queries sent to both of the clouds
Results are combined
Hybrid cloud tests (23206 | 23207)
CouchDB x 2
Tomcat x 2
Results - Hybrid Cloud 1
Results - Hybrid Cloud 2
Architecture - Federation creation
Client - Upload object 1
Client - Upload object 2
We presented a possible architecture for a hybrid cloud storage keeps sensitive data
, allowing to use
cheap other storages
at the same time
Researched different sharding scenarios, implemented one
Examining the confidence intervals and deviations lead us to conclude that response times show low deviations and stability. Single query takes 10 milliseconds, and in a hybrid cloud this amount is around 30 milliseconds, which are acceptable.
Multithreaded setup boosted our performance with a degree of ~3.
Several sharding scenarios
Several public cloud support instead of one
Exchangeable rules at runtime