Loading presentation...

Present Remotely

Send the link below via email or IM

Copy

Present to your audience

Start 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

Do you really want to delete this prezi?

Neither you, nor the coeditors you shared it with will be able to recover it again.

DeleteCancel

Make your likes visible on Facebook?

Connect your Facebook account to Prezi and let your likes appear on your timeline.
You can change this under Settings & Account at any time.

No, thanks

MSc Thesis Presentation

No description
by

Enver Bahar

on 30 October 2013

Comments (0)

Please log in to add your comment.

Report abuse

Transcript of MSc Thesis Presentation

"
Vision Cloud
(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
metadata
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!
- Amazon
- Windows Azure

- Cheaper
- Elastic


- Security?


hybrid
cloud
implementation
based on
content centric service

Prof. Michael Gerndt
Prof. Carsten Trinitis


Sebastian Dippl


Enver Bahar ( )

MSc Thesis
Contents
15-20 min
20-25 min
25 + min
5-15 min
0-5 Min
Introduction
Idea
Background
CCS
Hybrid Cloud
Architecture
Scenarios
Implementation
Results
Evaluation
Conclusion
Questions
Intro - Hybrid what?
Hybrid Cloud
implementation

based on

Content Centric Service
Background - Hybrid Cloud
A hybrid cloud environment is a composition of
two or more
clouds offering
benefits
of multiple deployment models
...
Background - Hybrid Cloud 2
private
cloud
public
cloud
Background - CCS 1
Private clouds
offer security at network level, and
public clouds
offer scalability and cost efficiency.
Hybrid cloud
is a combination of both of them
The approch of keeping some data
local
and using the
elasticity
of public cloud is much more practical for many business cases i.e.
Siemens Healthcare
and
Energy

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.
NEW HARDWARE
Option 1: Buy new hardware!
- Expensive and fixed costs
- Hard to maintain
- Administration costs


PUBLIC CLOUD
SOLUTION
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

Channel
Video ID
Title
Description
Duration
Category
Tags
Views
Comments
Likes
Dislikes
Rating
Favorites
Embeds

why not for all kinds
of data?
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
any
kind of storage thus preventing vendor lock-in
...
unified access
service layer



logic layer




storage access layer
Background - CCS 4
Background - CCS 5
Architecture - Sharding
How to distribute data between the clouds?

Sharding
is a type of database partitioning that separates very large databases the into smaller, faster, more easily managed parts

Different sharding mechanisms:
Vertical partitioning
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?
Data Federation
: A data structure that contains metadata about remote data rather than the actual data (
Cloud Federation
)

Components:
FederationInfo:
Holds metadata about clouds such as access credentials
FederationAdmin:
A management layer above the cloud systems and stores information(FederationInfo) about other clouds
FederationAdminUI:
Client side app enables a unified view among the clouds
ShardService:
Defines how data is sharded and queried

now
PUT

{
"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"
}

private
public
private
public
create object
{
"confidential" : "true"
}
private
public
create object
{
"confidential" : "false"
}
Results - Single Cloud 1
Youtube Deutsche Welle channel
700 mb
of metadata belonging to
46413
videos
Each video containing
20 metadata
fields
Made several number of queries at the same time
We sent
90
,
~2000
and
3000
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
Database:
CouchDB
Application Server:
Tomcat


Results - Single Cloud 2
Same dataset, divided into
23206
and
23207
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)
Database:
CouchDB x 2
Application Server:
Tomcat x 2


Results - Hybrid Cloud 1
Results - Hybrid Cloud 2
Results combined
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
Thank you.
Questions
goal
Architecture - Federation creation
Client - Upload object 1
Client - Upload object 2
Conclusion
Conclusion 3
We presented a possible architecture for a hybrid cloud storage keeps sensitive data
on premise
, 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.
Future work
Several sharding scenarios
Several public cloud support instead of one
Identity provider
Exchangeable rules at runtime



25
26
26
Full transcript