approaches for distributed information computation and processing
TRANSCRIPT
©2011 Nokia
Future Approaches for Distributed Information Computation and ProcessingSerguei BoldyrevPhD
Services & Software technologies inLocation & Commerce Technology evaluation
u-World Congress 2011
©2011 Nokia 2
Agenda• From Web of Information• To Web of Computing• Data• Computation• Managed Performance• Privacy• Conclusions
©2011 Nokia 3
The nature of the consumer’s relationships with companies is changing
From a monologue… to a conversation…into continuous relationships
From a unified… to a segmented…Into a dynamic personalized offering together with ecosystem
36
Microsoft
©2011 Nokia 4
Enhancing user experience
pastdevice only
presentdevice + service(s) + partners
futureSeamless User Experience across device + services + partner ecosystem
E72+Messaging
Nokia N950+Maps
Nokia 600 + Music
©2011 Nokia 5
Towards inclusive and sustainable ecosystem
Operators
People and Places (including Maps)Consumer understanding
Open Interfaces
OperatorServices
Attractive offering to mobile users Microsoft
Services NokiaServices
• Nokia targets 300M active Nokia Services users by the end of 2011 • 300M users of Cloud
• 800M Cloud-compatible devices exist in 2012• Cloud computing/energy/cost effective HW is on the vendor’s
roadmaps, 2013 and onwards• Cloud Computing subscribers would total nearly one billion by 2014
A sustainable ecosystem where Nokia services will be only one part of the offering
©2011 Nokia 6
From a Web of Information
People and Places (including Maps)Service Platforms
Open Interfaces
Share favorite songs with friends
Share content and micro-blog between
friends
Surface the most relevant contentLocate friends
around you on a map
• Cloud computing today serve email, apps, downloads and storage • Current "cloud apps" relying more on server-side processing• SaaS, PaaS, NaaS and other service models to emerge• Thin client, network as a computer, etc• Seamless information management between Client and Cloud
©2011 Nokia 7
Towards a Web of Computing
BetweenDevices & Infrastructure
Scale up/down to/from CloudBetweendevices
Within Operators & Infrastructures
©2011 Nokia 8
Why is this happening? (we agree with “Others”)
Because digital transformation of products and services are demanded by consumers, with increasing complexity driven by new customer
expectations
Multiple & continuous Interaction
Consumption based
Always on
Personalized and
custom
Transparent and
trusted
Everything as a service
Consumerexpectatio
n
©2011 Nokia 9
Ideal Scenario for the Future Solution• Mobile device is a true contributing participant , not just a depending client • Mobile device connects to multiple data sources (possibly independent clouds),
forming one logical Computation/Data and Service space• From any connected device, the user deals with the same data regardless of its
physical location
• Solution provides both data and distributed computing services• Solution access is managed by fine-grained security policies applied according to
the user’s context • From the design perspective, the Solution manages and hides complexity; does not
increase it• Solution data and services decrease development and maintenance costs• Solution data and service pricing model creates new money flows between players
©2011 Nokia 10
Opportunity of New Digital ExperienceWorld of fully deployed and available cloud infrastructures
The emergence of Distributed Systems that can seamlessly span the “information spaces” of multiple • hardware, • software, • information and • infrastructure providers
©2011 Nokia 11
Future Solution, technological disruptionsUtilizing the best parts of the Web development paradigm
• …Web-based applications, not Web sites…• and leveraging the special contextual capabilities of mobile devices (ex. WURFL databases on sensors)
Moving from a device-cloud and Web-centric models to a Hybrid model (Cloud & Edge computing)• Reuse (partition) information and computation in the Cloud, Infrastructure and expand it to devices• Elimination of special-purpose software to download, install applications, Service oriented infrastructure constructed as
a functional flows requested on demand• Enabling balancing of computation and relevant data between heterogeneous Cloud Back-Ends, Infrastructures (ex.
Spanner, HDFS, Web Intents etc) and devicesFostering faster, easier, richer application innovation and deployment through Cloud Back-End computation recycling
• Diverse Cloud Back-Ends can all leverage the infrastructure capabilities• Atomic computations are now deployed once to the Back-Ends and composed down to Infrastructure and clients, where
a set of functional flows or description of those form the actual service• Allowing Network Infrastructure to leverage data and computation workload, by taking care of services or provider of
services capabilities, distributing computation between Back-Ends and Infrastructure• Allowing more efficient contextual composition of services than purely device or Web-centric models
PlatformData
App
Device-centric model Web-centric model
ABC Cloud
DataPlatformApp
“Future” model
ABC Cloud
Computation & DataPlatform
App
XYZ Cloud
©2011 Nokia 12
1st Class Partner
Future Solution, Services interoperability
Edge
++
3 rd Party
Agnt
Regional DC
Master DCRegion
al DC
• A next generation distributed systems are dynamically spanned around virtual and physical infrastructures
• Service APIs are outsourced from Core DC to Cloud and beyond on-the-Edge
• Edge++ becomes a known and contributing node of the cloud infrastructure, service provisioning element
3rd Party
ODataAgn
tL&C
Cloud Service
Core
Edge
©2011 Nokia 13
New Digital Experience"My digital experience follows me regardless the computing environments around me“
• take advantage of finely granular and accountable processing services,
• integrate heterogeneous information sources, and
• ultimately free users of mundane challenges and details of technology usage
Greater reuse of information and computational tasks is just few steps away
©2011 Nokia 14
Opportunities and ChallengesDesign framework should be defined
Main aspects of • Data and Computation semantics, • Performance and Scalability of computation and
networking, • Threats to Security and Privacy
©2011 Nokia 15
Data - Common Data Model (CDM) enables portability of data across application domains
Policy Engine
Shared Data
IntegrationSemantics
Service1Logic
Service1 API
ClientA
ClientB
Service2Logic
Service2 API
Data API
Client Applications should not “own” data, instead, they should reflect the users’ desire to do something, to accomplish certain goals• to call someone, you need
Contacts, • to know what to do next, you need
Calendar, etc.
Data is still the same… • for example, it would be helpful for
Contacts to know who is the organizer of your next meeting
©2011 Nokia 16
Computation, the componentsClientA
ClientB
Backend EnvironmentClient API
Computation run-time
Environment
Agent 1
Agent 2
Agent 3
Client API
Computation run-time
EnvironmentRecycling
& Marshaling
Computations Store
Agent 4
Agent 5
Convenience APIPHP API Java API
Dist
ribut
ion Back end API
Dist
ribut
ion
Control/Ontologies
Computation
Dist
ribut
ion
To enable and create:• consistent (from
the semantics standpoint), and
• accountable components
Components that can be leveraged and reused in a larger, distributed information system
©2011 Nokia 17
Computation and Data perspectivePrerequisites
Core Data assets are encapsulated into results of Computation tasks
Objectives (but not limited to listed below)• Design and implementation of scenarios where computation can be described and represented in a system-wide understandable way
• enabling computation to be “migrated” (transmitted) from one computational environment to another
• such transmission would involve reconstruction of the computational task at the receiving end to match the defined semantics of task and data involved
• Construction of a system where larger computational tasks can be decomposed into smaller tasks or units of computation
• independent of what the eventual execution environment(s) of these tasks (that is, independent of hardware platforms or operating systems)
©2011 Nokia 18
Example
Computations and Data Store
Object <compute_1, compute_2, compute_3, …>
compute_3=filter(object)
context{object<compute_3>}
object=project(compute_3)
1. User Computations 6. Reconstruct computation
Computation set 3
Computation set 1
Computation set 2
Applications, User Context
Applications and Back End logic
3. Computation distributionput(compute_3)
5. Project computation to the Functional chain with User device context
get(compute_3)
4. Computation extraction
Users’ Computations [ ]
( ) { }
Granular and Reflective Process Representation
User Device
2. Selection of the context (functional chain)
Agent 3
Agent 1
lift’n’compose
Developer specifies ECMAScript bounded to the ComputationsBack End provides the
basic Computation sets
Agent 2
©2011 Nokia 19
Performance/Scalability, RequirementsThe new emerging paradigm of Distributed Systems requires an explicit orchestration of computation across
• Mobile Devices• the Edge, (thin elements of the Cloud “skin”)• Cloud and • the Core
Orchestration depends on components’ states and resource demand, through the proper exploitation of granular and therefore accountable mechanisms
©2011 Nokia 20
Scalable Computing
URIs
Computation environment
Computation environment
Legacy routines
Legacy routines
0xFFFF
0x0 0x0
0xFFFF … FFFF
Legacy routines
Distributed Computations based platform
Legacy routines
URIs
Front-End Side
Back-End SideDistribution/
sync
Com
puta
tions
SU
PERS
ET
Computations
subset
©2011 Nokia 21
Example
Computational branch x
Computation 1
Computation 2
Computation 3Connector
00
start
Each branch e.g. different nodes within Cloud/Infrastructure
Before Starts:Parameters to know beforehand (before start invoked):1) Capabilities2) Functional Flow specification (of how many parts we have in the map),
see legend in this slide (right).3) Cost functions (compound function): what are cost in order to bind next
Computation4) Rules (entity that drives this connector Computation and if not then it
directs elsewhere)
- capabilities
Cost function
Functional flow
Rules
e.g. Connector 00
e.g. Connector x0
Connector x0
Connector x1
Connector x2
=> First list the 1-4 parameters then proceed with load balancing phases
Computation 0
Connector x3
Computation x1
Within devices
Connector x1
Connector x2
Computation x1
Computation x2
Computation x2
Computation x3
Computation x3
Connector x3
Computation 4
stop
©2011 Nokia 22
A number of options• Granular latency control for diverse Data and Computational
load in hybrid architectures• Resolution of short term capacity decisions, including
• the determination of optimal configuration, • the number of (live) servers or • the migration of computation
• Resolution of long term capacity decisions • decisions concerning the development and • extension of data- and service architecture, choice of technology,
etc• Control of SLAs at different levels,
• e.g. consumer services, platform components, etc• Quality assessment of distributed software architecture
©2011 Nokia 23
Security, Identity, Privacy the placement and scope
Consumer Lifetime Value
(note that in this diagram, by “ontology” the semantics is meant)
©2011 Nokia 24
Privacy, the scopeTo provide mechanisms by which users can finely tune the • ownership, • granularity, • content, • semantics and • visibility of their data towards other parties
• Security controls whether a particular party has access to a particular item of data
• Privacy controls whether that data can or will be used for a particular purpose
• Security is very closely linked with privacy, however, tends to be about access control, whether an agent has access to a particular piece of data, whereas privacy is about constraints on the usage of data
©2011 Nokia 25
Distributed Data Infrastructure requirementsGiven any Distributed Data Infrastructure we require at minimum: • a security model to support the integrity of the data, • an identity model to denote the owner or owners of data
(and ideally also provenance of the data) and • an ontology to provide semantics or meaning to the data
Given these we can construct a privacy model where • the flow of data through the system can be monitored and • to establish boundaries where the control points for the data
can be placed
©2011 Nokia 26
Privacy Boundary• Inside the boundary• Outside the boundary• On the Edge
Implications for 3rd Parties– Compliance with standards– Ensuring compliance– Termination of contract
Application orService
Application orService
Hardware orO/S
UserInterface
otherApplication orService by IPC Application or
Service
otherApplication orService by IPC
Client
”Internet”
Back-End
©2011 Nokia 27
Conclusion - Driving agenda• Creation of tools and effective standards for the next developer and
consumer applications, • Computing platform for the Continues Service Delivery
o the fine-grained (and thus easily accountable) software frameworks developments
o anything as a service, e.g. giving hardware for “free”, just use the services• Backend infrastructure, backed up by application technologies to
enable qualitative leap in ICT industry offerings, from the perspectives of
o scalable technology and o multisided business models with high elasticity
Aiming and supporting future strategic growth areas and maintenance requirements
©2011 Nokia 29
References• ULS project• SlipStream project• Boldyrev, S., Kolesnikov, D., Lassila, O., & Oliver, I..
(2012). Data and Computation Interoperability in Internet Services. CLOSER.
©2011 Nokia
Thank you