- 1. Where P2P ? over the clouds into the clouds aside the
clouds?
- R. Minerva, A. Manzalini, C. Moiso
2. P2P Technology (a simple view)
-
- P2P system resources are often "free", coming from individuals
which volunteer their machines' CPUs, storage, and bandwidth
-
- P2P systems lack a centralized administrative entity that owns
and controls the peer resources. This makes it hard to ensure high
levels of availability and performance.
-
- Decentralized control also limits trust. Users can inspect the
memory and storage of a running application, meaning that
applications cannot safely store confidential information
unencrypted on peers. Nor can the application developer count on
any particular quantity of resources being dedicated on a machine,
or on any particular reliability of storage. These obstacles have
made it difficult to monetize p2p services. It should come as no
surprise that, so far, the most successful p2p applications have
been free, with Skype being a notable exception.
-
- The freeriding phenomenon (and the tragedy of commons)
Excerpts from http://berkeleyclouds.blogspot.com/ 3. P2P as a
complementary technology for Cloud Computing ?
- P2P techniques are often useful in building datacenter-scale
applications, including datacenter-scale applications that are
hosted in the cloud
- Actually the Key-Value Store techniques are often exploited by
big data centers
-
- Amazon Dynamo(see
http://www.allthingsdistributed.com/2007/10/amazons_dynamo.html)
Excerpts from http://berkeleyclouds.blogspot.com/ App 1 App 2
Appl m DISTRIBUTED APPLICATIONS Node 1 Node 2 Node n DISTRIBUTED
HASH TABLE put (H(key), value); get (H(key), data); H(Key)is
ahashing function of the Key 4. Or as an Alternative ? P2P Range of
Applicability Distributed Computing File Sharing Collaboration
Platforms Communication DataStorage Search Engine Dark nets
P2P Systems 5. Are Operators out of the emerging trend of the
(Data) Service/Platform/Infrastructure as a Service ? 6. The
Prevalent Paradigm For Services Is Client - Server
- Client Server paradigm is used in Web Interactions
- New Services are taking advantage of this consolidated
technology
-
- Rich Internet Application
- But ...Can the Data Centers grow forever?
-
- Google maintains over 450,000 servers (estimates)
-
- http://www.tnl.net/blog/2004/04/30/ how-many-google-machines
/
-
- Google Data Centers: $3,000 A Square Foot?
-
-
http://www.datacenterknowledge.com/archives/2007/11/12/google-data-centers-3000-a-square-foot/
-
- Estimates of the power required for over 450,000 servers range
upwards of 20megawatts , which cost on the order of US$2 million
per month in electricity charges
Web Client (Browser) Web Client (Browser) Web Client (Browser)
Internet Web Server http http http http 7. P2P Computing as a
competitor of Client Server?Capacity (HD System) = (b i , s i , f i
, p i ) Where: b i= bandwidth of node i s i= storage of node i f i=
files of node i p i= processing of node i Capacity (Centralized
System) = {b S , s S , f S , p S } Where: b S= bandwidth of the
Server System s S= storage of the Server System f S= files stored
in the Sever System p S= processing in the Server System
Optimization is a problem Optimization is a function of a business
model 8. www.gapinvoid.com A Few Issues 9. Example of Projects
trying to overcome P2P Cons ... The
nanodatacenter(http://www.nanodatacenters.eu/) 10. Autonomic
Computing
- What is it? (from wikipedia)
-
- IBM has defined the following four functional areas:
- Autonomic Networking follows the concept ofAutonomicComputing .
Its ultimate aim is to apply autonomic concepts to Networks.
- Functions of an Autonomic Network:
-
- Autognostic, Configuration management, Policy management,
Autodefense, Security
-
- Connection Fabric (i.e., The connection fabric supports the
interaction with all the elements and sub-systems of the autonomic
system)
- Principles of Autonomic Networking
-
- Compartmentalization (how to implement the operational rules
and administrative policies for a given communication context)
http://www-03.ibm.com/servers/eserver/zseries/zos/sm/autonomic.html
Why? Millions of pervasive nodes!! There is the need of a new
approach Users should be bothered by such a cumbersome task! 11.
Example Of Waste Of Resources
- Actually a p2p Overlay Network is considered
infrastructure-less, but ...
-
- In reality it is using the network infrastructure ...
-
- E.g., two nodes seen as adjacent by the p2p network can be very
far each other. T-Lab suggests mechanisms to improve the
situation
- There is the need to manage the network resources in order to
give a fair service to ALL the users.
- There is the need to cooperate
- Needs for interoperability: means definition of common
protocols and functions
Overlay Net See for instance
http://www.cambridge-mit.org/project/home/default.aspx?objid=2313
P4P and IETF ALTO are trying to solve these problems 12.
Application of Gossiping Algorithm to FCAPS areas(courtesy of
CASCADAS)
- fault detection and recovery
- for each contract of type T there is a back-up contract, to be
used if B fails;
- the nodes in achieving(T) updates the back-up contract in case
B fails;
- the nodes in contracting(T) help A in case B and B fails;
- when B is under-loaded, B informs all its neighbors in
achieving(T);
- if B, one of the neighbors of B, is overloaded, it answers to B
by accepting the offer;
- when A, with an active contract with B, sends a new request, B
informs A to redirect its contract to B;
- Usable as thestand-by algorithm for power saving
13. Simulation snapshots (from Cascadas) State ofNodeswith
distributed LB Legenda: Nodesavailable to get traffic Nodesin
normal load Nodesin overload State ofNodeswithout LB Thanks to A.
Manzalini, C. Moiso 14. Cross Layer Design: The End Of The OSI
Model??http://www.uni-marburg.de/fb12/verteilte_systeme/forschung/adhoc_routing_emul
- Wireless Networks are pushing for a closer cooperation of
functions in order to better respond to a fast changing
communication environment:
-
- Reflective middleware(able to adapt)
-
- Cross Layer Design(ability to get the relevant information from
any OSI layer)
-
- Protocol dependence of the middleware
-
- Cognitive Layer(it can perceive current network conditions, and
then plan, decide and act on those conditions)
15. A glimpse to the future ...
- DirecNet: A group of industry players is proposing
anopen-standarddirectional networking system designed to
provide1-gigabit-per-second data communicationwith anyone in a
network on the ground, in the air or at sea,within hundreds of
miles . The plannedmobile mesh network , known as DirecNet, will
use a network-friendly data link waveform to provide secure
Internet Protocol communications up to 1 gbps for all types of air
and surface platforms. DirecNet would use fast-steered directional
antennas to substantially boost link power and operating range, and
to permitreuse of radio frequency (RF) spectrum .Any DirecNet node
can serve as a relay . This will multiply connectivity and extend
the range to beyond line of sight.(
http://www.military-information-technology.com/print_article.cfm?DocID
=1539 )
-
- The DirecNet Task Force anticipatesthe completion of the
specification within 2 years(
https://www.opengroup.org/conference-live/uploads/40/11007/Forum_Reports.pdf#search=%22goals%20direcnet%20military%22
)
16. Leveraging the TELCO infrastructure www.gapinvoid.com 17.
Incipit
- Positive Dilemmas : Network Operators should investigate
disruptive scenarios for Future Communications and Services in
order to be aware of how to transform their biz models and exploit
network and service infrastructures;
- One of the most challenging scenario is characterized by:
-
- High pervasiveness (computing systems distributed and connected
everywhere);
-
- Connectivity as a commodity;
-
- Structure as Network of Networks (adaptive networks);
-
- Intelligence at the edges (terminals and end points);
- This scenario is highly disruptive as it implies moving
intelligence at the edge of the network reinforcing the end-to-end
argument and an open environment for services;
- There is a need of understanding:
-
- The potential evolution of this scenario (in terms of
architectures, technologies and services);
-
- Which is the role of the Network Operator (biz model,
architectures, technologies, );
18. Architectural vision and principles All rights reserved 19.
Next: Open Questions
- Which mix of networking and computing paradigms and
technologies will enable the scenario (characterized by high
pervasivity, with connectivity as a commodity and Network of
Networks) ?
- How to optimize allocation, sharing and management of pervasive
resources, to guarantee the best Users experience whilst reducing
the systems and Operators costs ?
- How to ensure the sustainability of the entire ecosystem, to
avoid Tragedy of Commons due to unplanned usage of shared resources
?
- How to create, nurture and protect an ecosystem ?
20. Further contacts
http://www.gapingvoid.com/Moveable_Type/archives/2006_03.html