Transcript Document

A Survey on Network Storage
Ning Zong
Haibin Song
Richard Alimi
Richard Yang
Network Storage System Components
Client (2) Access Protocol
(1) Discovery
(3) Operations
(4) Access & Resource
Control
(5) Storage
Mode
(6) Operational Model
Network Storage
NFS (Network File System)


NFS allows a user on a client
computer to access files over a
network in a manner similar to
how local storage is accessed.
pNFS allows clients to access
storage devices directly and in
parallel.
NFS Server
NFS Client
NFS Client
NFS Client
 Doesn’t provide resource control (bandwidth and connections)
 Access control with many P2P peers is difficult
Web Cache


Deployed between web servers and
clients to save copies of server
responses for future reuse to reduce
latency and network traffic.
HTPT: Proposal to share chunks of
P2P files/streams using HTTP protocol
with cache-control headers.
web Client
Req (URL_1)
Req (URL_2)
web
cache
web Client
Req (URL_1)
web Server
Resp (Page_A)
Resp (Page_B)
web Server
Resp (Page_A)
web Client
 Doesn’t provide resource control (bandwidth and connections)
 Access control with many P2P peers is difficult
Transparent P2P Cache





Allows duplicate P2P content to be cached and supplied locally
Impersonates P2P peer → requires support of P2P application
protocols
Another option: modify peer list to include cache
Upstream and downstream traffic reduction
Use of DPI means no discovery needed by application
 Doesn’t provide resource control (bandwidth and connections)
 Cache is complicated due to the evolving P2P applications
Non-Transparent P2P Cache


Allows duplicate P2P content
to be cached and supplied
locally
Operates as traditional P2P
peer → requires support of
P2P application protocols
 Doesn’t provide resource control (bandwidth and connections)
 Cache is complicated due to the evolving P2P applications
CDNs (Akamai, Limelight,
CloudFront)



Allows content to be
distributed globally
Push contents to network edge
Lower latency, reduces
upstream bandwidth
requirement by content owner
 Doesn’t provide resource control (bandwidth and connections)
 Access control with many P2P peers is difficult
Amazon S3



Online storage service
Users create buckets; buckets
contain stored objects
Popular backend storage for
other services
 Doesn’t provide resource control (bandwidth and connections)
Windows Azure


Storage for distributed (cloud)
applications
BLOB Service API supports
storage of binary data
 Doesn’t provide resource control (bandwidth and connections)
 Access control with many P2P peers is difficult
OceanStore



Globally distributed storage
Multiple storage providers pool
resources together
Focus on resilliency
 Doesn’t provide resource control (bandwidth and connections)
 Doesn’t support explicit control over location of data in the network
ISCSI (Internet Small Computer
System Interface)


Protocol for carrying SCSI
commands over IP
Popular for Storage Area
Networks
 Doesn’t provide resource control (bandwidth and connections)
 Access control with many P2P peers is difficult
Q&A
@$%#^*&%
%^&#$
*&^$%
^
!@!@^
&%%$