Introduction to Machine Learning Unit-3 for II MECH
Ā
CHARM(A Cost-Efficient Multi-Cloud Data Hosting Scheme with High Availability)
1. Introduction
Multi-Cloud Scenario
Motivation
Available Data Hosting Schemes
The proposed Scheme CHARM
Conclusion and Future Scope
CHARM
(A Cost-Eļ¬cient Multi-Cloud Data Hosting
Scheme with High Availability)
Deeksha Arya
deekshasheokand@gmail.com
Computer Engineering Department
NIT Kurukshetra
May 5, 2016
Deeksha Arya CHARM: Multi-Cloud Data Hosting Scheme
2. Introduction
Multi-Cloud Scenario
Motivation
Available Data Hosting Schemes
The proposed Scheme CHARM
Conclusion and Future Scope
Outline
1 Introduction
Cloud Computing
2 Multi-Cloud Scenario
Single Cloud vs Multi-Cloud
Problems in Single Cloud Data Hosting
Problems in Multi-Cloud Data Hosting
3 Motivation
4 Available Data Hosting Schemes
Replication
Erasure Coding
5 The proposed Scheme CHARM
Architecture
Algorithms Used
6 Conclusion and Future Scope
Deeksha Arya CHARM: Multi-Cloud Data Hosting Scheme
3. Introduction
Multi-Cloud Scenario
Motivation
Available Data Hosting Schemes
The proposed Scheme CHARM
Conclusion and Future Scope
Cloud Computing
What is Cloud Computing?
Cloud Computing is the use of computing resources(hardware
and software) that are delivered as a service over a network.
Deeksha Arya CHARM: Multi-Cloud Data Hosting Scheme
4. Introduction
Multi-Cloud Scenario
Motivation
Available Data Hosting Schemes
The proposed Scheme CHARM
Conclusion and Future Scope
Cloud Computing
How does it work?
In cloud computing, users access the data, applications or any
other services with the help of a browser regardless of the device
used and the userās location.
Deeksha Arya CHARM: Multi-Cloud Data Hosting Scheme
8. Introduction
Multi-Cloud Scenario
Motivation
Available Data Hosting Schemes
The proposed Scheme CHARM
Conclusion and Future Scope
Single Cloud vs Multi-Cloud
Problems in Single Cloud Data Hosting
Basic Principle of Multi-cloud Data Hosting
Problems in Multi-Cloud Data Hosting
Single Cloud vs Multi Cloud
Which one is more beneļ¬cial?
How to decide?
Who will decide?
What if the situation changes?
Deeksha Arya CHARM: Multi-Cloud Data Hosting Scheme
9. Introduction
Multi-Cloud Scenario
Motivation
Available Data Hosting Schemes
The proposed Scheme CHARM
Conclusion and Future Scope
Single Cloud vs Multi-Cloud
Problems in Single Cloud Data Hosting
Basic Principle of Multi-cloud Data Hosting
Problems in Multi-Cloud Data Hosting
Heterogeneous Clouds
Existing clouds exhibit great heterogeneities in terms of both
working performances and pricing policies.
For instance,
Google Cloud Storage charges more for bandwidth
consumption,
Amazon S3 charges more for storage space
Deeksha Arya CHARM: Multi-Cloud Data Hosting Scheme
10. Introduction
Multi-Cloud Scenario
Motivation
Available Data Hosting Schemes
The proposed Scheme CHARM
Conclusion and Future Scope
Single Cloud vs Multi-Cloud
Problems in Single Cloud Data Hosting
Basic Principle of Multi-cloud Data Hosting
Problems in Multi-Cloud Data Hosting
Vendor Lock-in Risk
Customers usually put their data into a single cloud and then
simply trust to luck. This is subject to the so-called vendor
lock-in risk, because customers would be confronted with a
dilemma if they want to switch to other cloud vendors.
For example, moving 100 TB of data from Amazon
S3(California data-center)to Aliyun OSS (Beijing data-center)
would consume as much as 12,300 (US) dollars.
Deeksha Arya CHARM: Multi-Cloud Data Hosting Scheme
11. Introduction
Multi-Cloud Scenario
Motivation
Available Data Hosting Schemes
The proposed Scheme CHARM
Conclusion and Future Scope
Single Cloud vs Multi-Cloud
Problems in Single Cloud Data Hosting
Basic Principle of Multi-cloud Data Hosting
Problems in Multi-Cloud Data Hosting
The vendor lock-in risk makes customers suļ¬er from price
adjustments of cloud vendors.
Another issue is availability.
Clearly, it is unwise for an enterprise or an organization to
host all data in a single cloudā āyour best bet is probably not
to put all your eggs in one basket ā.
Deeksha Arya CHARM: Multi-Cloud Data Hosting Scheme
12. Introduction
Multi-Cloud Scenario
Motivation
Available Data Hosting Schemes
The proposed Scheme CHARM
Conclusion and Future Scope
Single Cloud vs Multi-Cloud
Problems in Single Cloud Data Hosting
Basic Principle of Multi-cloud Data Hosting
Problems in Multi-Cloud Data Hosting
Multi-Cloud Scenario
Basic Principle of Multi-cloud Data Hosting
Deeksha Arya CHARM: Multi-Cloud Data Hosting Scheme
13. Introduction
Multi-Cloud Scenario
Motivation
Available Data Hosting Schemes
The proposed Scheme CHARM
Conclusion and Future Scope
Single Cloud vs Multi-Cloud
Problems in Single Cloud Data Hosting
Basic Principle of Multi-cloud Data Hosting
Problems in Multi-Cloud Data Hosting
Problems in Multi-Cloud Data Hosting
Two critical problems:
1 How to choose appropriate clouds to minimize monetary cost
in the presence of heterogeneous pricing policies?
2 How to meet the diļ¬erent availability requirements of diļ¬erent
services?
Deeksha Arya CHARM: Multi-Cloud Data Hosting Scheme
14. Introduction
Multi-Cloud Scenario
Motivation
Available Data Hosting Schemes
The proposed Scheme CHARM
Conclusion and Future Scope
Motivation
Major concern for customers: Which cloud(s) are suitable
for storing their data and what hosting strategy is
cheaper?
CHARM-the proposed Data Hosting scheme helps the
customers to select several suitable clouds and an appropriate
redundancy strategy to store the data with
1. Minimum monetary cost and
2. Guaranteed Availability
In addition, it also provides the function of re-distributing
the data according to the variations of data access patterns
and pricing of clouds.
Deeksha Arya CHARM: Multi-Cloud Data Hosting Scheme
15. Introduction
Multi-Cloud Scenario
Motivation
Available Data Hosting Schemes
The proposed Scheme CHARM
Conclusion and Future Scope
Replication
Erasure Coding
Available Data Hosting Schemes
Available Redundancy Mechanisms for Data Hosting:
1. Replication
2. Erasure Coding
Diļ¬erent schemes uses these redundancy mechanisms in
diļ¬erent ways( Random/Greedy).
Deeksha Arya CHARM: Multi-Cloud Data Hosting Scheme
16. Introduction
Multi-Cloud Scenario
Motivation
Available Data Hosting Schemes
The proposed Scheme CHARM
Conclusion and Future Scope
Replication
Erasure Coding
Replication
Traditional Storage Mode: Put the replicas into three
diļ¬erent storage nodes.
Advantages:
Provides High Availability
The data is lost only when the three nodes all crash.
Dis-advantage: Occupies 2x more storage space.
Deeksha Arya CHARM: Multi-Cloud Data Hosting Scheme
17. Introduction
Multi-Cloud Scenario
Motivation
Available Data Hosting Schemes
The proposed Scheme CHARM
Conclusion and Future Scope
Replication
Erasure Coding
Erasure Coding
Erasure coding is proposed to reduce storage consumption
greatly while guaranteeing the same or higher level of data
reliability as that of Replication.
Reed-Solomon code - A representative erasure-coding scheme:
Data of a segment is encoded into n blocks including
1. m data blocks and,
2. n-m coding blocks
These blocks are put into n diļ¬erent clouds.
Data can be restored using any m chunks in the segment.
Deeksha Arya CHARM: Multi-Cloud Data Hosting Scheme
18. Introduction
Multi-Cloud Scenario
Motivation
Available Data Hosting Schemes
The proposed Scheme CHARM
Conclusion and Future Scope
Replication
Erasure Coding
Erasure Coding
Advantages:
Data availability can be guaranteed with lower storage space
(compared with replication).
Up to n ā m simultaneous chunk failures can be tolerated.
Dis-advantages:
A read access has to be served by multiple clouds that store
the corresponding data blocks.
Consequently, erasure coding cannot make full use of the
cheapest cloud as what replication does.
Deeksha Arya CHARM: Multi-Cloud Data Hosting Scheme
19. Introduction
Multi-Cloud Scenario
Motivation
Available Data Hosting Schemes
The proposed Scheme CHARM
Conclusion and Future Scope
Replication
Erasure Coding
Replication in terms of Erasure Coding
Replication is a special case of Erasure Coding with m=1.
That is, in Replication we have
m = 1 data block
n ā m = n ā 1 coding blocks and
Up to n ā m = n ā 1 simultaneous failures can be tolerated.
Deeksha Arya CHARM: Multi-Cloud Data Hosting Scheme
20. Introduction
Multi-Cloud Scenario
Motivation
Available Data Hosting Schemes
The proposed Scheme CHARM
Conclusion and Future Scope
Architecture
Algorithms Used
CHARM
CHARM stands for a Cost-eļ¬cient data Hosting model with
high Availability in heteRogenous Multi-cloud system.
CHARM combines the two widely used redundancy
mechanisms, i.e., replication and erasure coding.
CHARM ā” (Replication + Erasure Coding)
Deeksha Arya CHARM: Multi-Cloud Data Hosting Scheme
22. Introduction
Multi-Cloud Scenario
Motivation
Available Data Hosting Schemes
The proposed Scheme CHARM
Conclusion and Future Scope
Architecture
Algorithms Used
Multi-Cloud Scenario
Basic Principle of Multi-cloud Data Hosting
Deeksha Arya CHARM: Multi-Cloud Data Hosting Scheme
23. Introduction
Multi-Cloud Scenario
Motivation
Available Data Hosting Schemes
The proposed Scheme CHARM
Conclusion and Future Scope
Architecture
Algorithms Used
Components of CHARM
There are four main components in CHARM:
1. Data Hosting
2. Storage Mode Switching (SMS)
3. Workload Statistic
4. Predictor
Deeksha Arya CHARM: Multi-Cloud Data Hosting Scheme
24. Introduction
Multi-Cloud Scenario
Motivation
Available Data Hosting Schemes
The proposed Scheme CHARM
Conclusion and Future Scope
Architecture
Algorithms Used
Working of the Components
1. Data Hosting stores data using replication or erasure coding,
according to the size and access frequency of the data.
2. Storage Mode Switching (SMS) decides whether the
storage mode of certain data should be changed from
replication to erasure coding or in reverse, according to the
output of Predictor.
3. Workload Statistic keeps collecting and tackling access logs
to guide the placement of data. It also sends statistic
information to Predictor which guides the action of SMS.
4. Predictor is used to predict the future access frequency of
ļ¬les.
Deeksha Arya CHARM: Multi-Cloud Data Hosting Scheme
27. Introduction
Multi-Cloud Scenario
Motivation
Available Data Hosting Schemes
The proposed Scheme CHARM
Conclusion and Future Scope
Architecture
Algorithms Used
Key Points
1. What should be the value of n? (n varies from 2 to nmax )
2. Which n clouds should be selected?
3. What should be the value of m? (m varies from 1 to n )
4. Goals:
Minimize the cost
Maximize the availability
5. Complexity of the Algorithm: O(NLogN)
Deeksha Arya CHARM: Multi-Cloud Data Hosting Scheme
29. Introduction
Multi-Cloud Scenario
Motivation
Available Data Hosting Schemes
The proposed Scheme CHARM
Conclusion and Future Scope
Conclusion and Future Scope
CHARM makes ļ¬ne-grained decisions about which storage
mode to use and which clouds to place data in and hence
helps the user to distribute the data on diļ¬erent clouds in a
cost eļ¬ective manner.
This system can be enhanced by developing an automatic
update methodology that updates only the required data
blocks.
The enhanced version will save the time for downloading and
uploading the ļ¬le again.
Deeksha Arya CHARM: Multi-Cloud Data Hosting Scheme
30. Introduction
Multi-Cloud Scenario
Motivation
Available Data Hosting Schemes
The proposed Scheme CHARM
Conclusion and Future Scope
References
Quanlu Zhang, Shenglong Li, Zhenhua Li, Yuanjian Xing, Zhi Yang, and
Yafei Dai (2015)
CHARM: A Cost-Eļ¬cient Multi-Cloud Data Hosting Scheme with High
Availability
IEEE TRANSACTIONS ON CLOUD COMPUTING, VOL. 3, NO. 3,
372-386, JULY-SEPTEMBER 2015.
Gade Pooja V., Mate Shweta, A. Janjalkar Priti S. (2015)
Multi-Cloud Hosting for Performance Optimization and Security
International Journal for Scientiļ¬c Research and Development, VOL. 3,
Issue 8 ā ISSN (online), 2321-0613 2015.
Deeksha Arya CHARM: Multi-Cloud Data Hosting Scheme