下載/瀏覽Download

Download Report

Transcript 下載/瀏覽Download

Research of P2P Architecture based
on Cloud Computing
Speaker :
吳靖緯 MA0G0101
2011.8.5
Outline
• CLOUD COMPUTING ARCHITECTURE AND
IMPLEMENTATION BASED ON P2P
• ANALYSIS AND EVALUATION
• CONCLUSION AND FUTURE WORK
CLOUD COMPUTING ARCHITECTURE AND
IMPLEMENTATION BASED ON P2P
A. P2P Cloud Computing architecture
• Cloud Computing architecture based on P2P includes three
basic roles: User, Central Peer and Side Peer.
• As shown in Fig. 1. Central Peer and Side Peer constitute two
peer network, respectively called Central P2P Network and
Side P2P Network.
Fig. 1 The cloud computing architecture based on P2P
B. Fault-tolerant Mechanisms of MapReduce based on P2P
• Fault-tolerant recovery strategy in Cloud Computing
architecture based on P2P is implemented through the
following steps:
a) Users can query the primary server to get the available server
list which is listed according to each server workload.
b) C1 act as the primary node. C2 and C3 are notified by C1
and act as the backup node.
c) C1 queries peers of Side P2P Network to get the list of the
available peers and selects these peers of the smallest
workload to process the current task.
d) When the primary peer C1 is detected by the backup nodes
to have an error, the backup peer C2 is activated to execute
the task of the primary peer.
e) The new primary peer can re-find a new backup peer in the
rest of the peers to satisfY the original backup node number.
f) The task is re-executed in the primary peer.
C. File Storage Mechanism based on P2P
• The paper uses DHT technology to achieve P2P network data
storage and location and address the structure distributed
storage problems.
• Peer Servers of Side P2P Network has three functional module:
Index Module, Route Module and Data Module.
ANALYSIS AND EVALUATION
• For the P2P storage system, the search time is reduced because
metadata is backed up in the P2P network.
• The system can avoid to the search error of the peers failure.
• P2P network has very good scalability, the scale of the system
can be expanded according to the needs of users.
CONCLUSION AND FUTURE WORK
• This article fully uses the advantages of P2P network to build a
cloud computing architecture.
• This whole system have good usability, scalability and fault
tolerance.
• The next works include
1) Constructing the P2P model based on MapReduce which
achieve distributed computing.
2) researching data storage and replication strategy in the
architecture.