Conference PaperPDF Available

NOVI Tools and Algorithms for Federating Virtualized Infrastructures

Authors:

Abstract and Figures

The EC FP7/FIRE STREP project NOVI - Network Innovation over Virtualized Infrastructures - explores efficient approaches to compose virtualized e-Infrastructures towards a holistic Future Internet (FI) cloud service. Resources belonging to various levels, i.e. networking, storage and processing are in principle managed by separate yet inter-working providers. In this ecosystem NOVI aspires to develop and validate methods, information systems and algorithms that will provide users with isolated slices, baskets of resources and services drawn from federated infrastructures. Experimental research accomplished thus far concludes the first phase of NOVI, with early prototypes of semantic-aware advanced control & management plane components being deployed and tested. The NOVI testing environment is based on combining PlanetLab and FEDERICA, two dissimilar virtualized experimental infrastructures with attributes widely anticipated in a FI cloud. This federated testbed is stitched at the data plane via the NSwitch, a distributed virtual switch developed within NOVI.
Content may be subject to copyright.
F. Álvarez et al. (Eds.): FIA 2012, LNCS 7281, pp. 213–224, 2012.
© The Author(s). This article is published with open access at SpringerLink.com
NOVI Tools and Algorithms for Federating Virtualized
Infrastructures
Leonidas Lymberopoulos1, Mary Grammatikou1, Martin Potts2, Paola Grosso3,
Attila Fekete4, Bartosz Belter5, Mauro Campanella6, and Vasilis Maglaris1
1 National Technical University of Athens
2 Martel Consulting
3 Universiteit van Amsterdam
4 Eötvös Loránd Tudományegyetem
5 Poznan Supercomputing and Networking Center
6 Consortium GARR
Abstract. The EC FP7/FIRE STREP project NOVI - Network Innovation over
Virtualized Infrastructures - explores efficient approaches to compose
virtualized e-Infrastructures towards a holistic Future Internet (FI) cloud
service. Resources belonging to various levels, i.e. networking, storage and
processing are in principle managed by separate yet inter-working providers. In
this ecosystem NOVI aspires to develop and validate methods, information
systems and algorithms that will provide users with isolated slices, baskets of
resources and services drawn from federated infrastructures. Experimental
research accomplished thus far concludes the first phase of NOVI, with early
prototypes of semantic-aware advanced control & management plane
components being deployed and tested. The NOVI testing environment is based
on combining PlanetLab and FEDERICA, two dissimilar virtualized
experimental infrastructures with attributes widely anticipated in a FI cloud.
This federated testbed is stitched at the data plane via the NSwitch, a distributed
virtual switch developed within NOVI.
Keywords: NOVI, Future Internet, FIRE, Virtualization, Federation.
1 The NOVI Project: Goals and Objectives
We report in this paper experimental work within the NOVI FIRE project [1] towards
a semantic-aware control and management plane for federating heterogeneous
virtualized infrastructures and for establishing data plane connectivity amongst virtual
resources offered by separate virtualized infrastructures. The goal is to offer
automated advanced capabilities to users of the federation: intelligent resource
mapping, policy-driven access and resource allocation, context aware resource
discovery, transparent data plane connectivity and monitoring of combined user slices
and substrate resources across domains. Experimental tool development and
validation are performed in a testbed environment, based on two dissimilar virtualized
infrastructures: FEDERICA [2] and PlanetLab [3]. The former, partially supported by
the EC FP7/Capacities Programme, provides users with a combination of Virtual
214 L. Lymberopoulos et al.
Machines and Logical Routers, interconnected via Layer 2 VLAN technology
extended over dedicated circuits provisioned by European National Research &
Education Networks (NRENs) and GÉANT [4]; the latter is a popular experimental
infrastructure, partially supported by the US NSF GENI Programme [5], that offers
collections (slices) of virtual computing resources (slivers) within more than a
thousand hosts, globally distributed over the legacy Internet. The selection of these
platforms provides NOVI with a combined testbed, exposing experiments to a wide
range of attributes as expected in a Future Internet federated cloud: FEDERICA
combines commercial virtualization tools providing virtual machines and logical
routers, interconnected with gigabit controlled connectivity; PlanetLab offers a highly
distributed virtual machine selection interconnected over the existing public Internet,
thus presenting distributed applications experiments with actual connectivity
limitations and unpredictable wide area networking behavior. In conclusion, the
combined PlanetLab and FEDERICA testbed for NOVI’s experimental research
captures basic features applicable in federated heterogeneous environments that are
expected to serve a wide range of user communities.
The paper is organized as follows. Section 2 presents our work on a domain-
independent Information Model aiming to capture the main abstractions of shared
resources and services within a NOVI federation. Section 3 presents the main
components of NOVI’s federated control and management plane and provides an
overview of NOVI’s distributed virtual switch (NSwitch) for data plane stitching.
Section 4 presents the combined testbed for NOVI’s prototype deployment and
experimentation. Finally, section 5 concludes the paper and provides directions for
future work in the remainder of the project.
2 NOVI Information Model
An agreed-upon Information Model (IM) provides consistent and shared semantics
and descriptions of available resources and services in a federated environment. In
NOVI we developed a novel IM and the associated data models as existing IM efforts,
listed in NOVI Public Deliverable D2.1: Information Models for Virtualized
Architectures [6] did not cover our two-fold objective: (a) to support the modeling
abstractions to cater for a federation of infrastructures, e.g. the FEDERICA and
PlanetLab platforms of the NOVI’s testbed; (b) to include the necessary concepts so
that can be used to model other Future Internet (FI) infrastructures that could
participate in a NOVI-like federation.
We fully embraced a Semantic Web approach and defined data models using the
Web Ontology Language - OWL [7]. This choice has been driven by the desire to
support reasoning and context awareness, which in turn allow NOVI to create
efficient and complex services with resources available within the federation.
The NOVI IM consists of three distinct but related ontologies; this modular
approach is chosen on purpose to make the model more easily usable outside the
project by parties interested in specific aspects. The NOVI IM defines a resource
ontology, a monitoring ontology and a policy ontology.
The Resource Ontology provides the concepts and methods to describe the
resources offered by Future Internet platforms and how they are connected together in
a federated environment. This ontology provides the basis for topology and request
NOVI Tools and Algorithms for Federating Virtualized Infrastructures 215
descriptions and the terminology for describing physical nodes, virtual nodes, virtual
topologies, etc. The Resource Ontology supports the operation of all the services of
NOVI’s Federated Control & Management Architecture, which will be presented in
the Section 3 of this paper. For example, it is used to express requests within the
NOVI GUI or by the Resource Information Service and the Intelligent Resource
Mapping Service to communicate when coordinating the exchange of information
about resources suitable for the embedding of virtual resources. The Monitoring
Ontology extends the Resource Ontology to provide descriptions of the concepts and
methods of monitoring operations, such as details about monitoring tools, how these
relate to the resources, types of measurements that can be gathered etc. This ontology
provides the primary support to the operation of the Monitoring Service. Finally, the
Policy Ontology also extends the Resource Ontology by providing descriptions of the
concepts and methods for the management and execution of policies defined within
member platforms of a NOVI federation. This ontology supports the operation of the
Policy Service. More information on the developed ontologies can be found in the
project’s public deliverable D2.2: First Information & Data Models [8].
3 NOVI Federated Data, Control and Management Plane
Architecture
NOVI’s novel algorithms, methods and services are initially based on the Slice
Federation Architecture - SFA [9] as developed for the PlanetLab control &
management plane federation. In SFA, a resource specification - RSpec is an XML-
file describing resources bound and available to a user slice in terms of hardware
characteristics, network facilities, constraints and dependencies on their allocation.
NOVI extends SFA with advanced context-aware federation mechanisms (intelligent
resource allocation, monitoring, policy management and virtualized resources
discovery) and automating slice control & management operations anticipated within
a complex NOVI federation.
The high level overview of the NOVI Data, Control & Management (C&M)
architecture is shown in Fig. 1. It consists of three different layers:
1. At the bottom layer heterogeneous platforms (domains, infrastructures) contain
the virtual resources to be allocated to user requests for combined slices. Data
plane connectivity within a NOVI federated slice is achieved using NOVI’s
Distributed Virtual Switch – NSwitch
2. The middle layer components are used to provide basic C&M federation
capabilities across platforms. In the figure we depict implementation choices
referring to SFA (e.g. cross-domain authentication via synchronized registries
and user-specified slice operations)
3. The top layer implements NOVI C&M services that aim at offering advanced
capabilities to the federation users (e.g. intelligent resource mapping, policy-
driven access and resource allocation, context aware resource discovery,
transparent monitoring of combined user slices and substrate resources across
domains). It leverages federation mechanisms of the middle layer (SFA),
complementing them with advanced C&M functionality.
216 L. Lymberopoulos et al.
Fig. 1. NOVI Federated Data, Control & Management Architecture
For each infrastructure (platform) in the federation, as demonstrated within the
NOVI testbed, separate NSwitch, SFA and NOVI C&M instances need to be
deployed. In what follows we outline functionality of components within the latter.
3.1 NOVI API
The NOVI API provides the entry point for interacting with NOVI C&M services. It
has three main tasks: (1) Accept requests from authenticated users containing
resources requirements represented in NOVI Information model; (2) Handle and
deliver the request to the appropriate component within NOVI Service Layer; (3)
Provide user feedback on how their request is handled before the experiment starts
being executed in its related NOVI slice.
NOVI Tools a
n
As shown in Fig.2, the N
O
based on the Ontology Insta
n
in the development phase
o
users to create and send req
u
GUI provides an intuitive dr
a
users to define relations be
t
virtual network topology al
o
request, the GUI generates a
n
NOVI API by means of an
H
Using the NOVI GUI
(
can choose from the avail
a
for his experiment.
3.2 Resource Informat
The
R
esource Information
NOVI services to acquir
e
resources. Resource disc
o
across the federated virtu
a
scalable query process.
R
underlying platform, to res
e
uses the Monitoring Servic
e
and the Policy Service to g
e
RIS exploits the feature
s
of resource discovery and t
o
uses a database engine bas
e
resource selection. The da
t
The RIS uses the AliBaba t
Java objects that describe
software components of th
e
3.3 Intelligent Resourc
The Intelligent Resource
M
user requests for virtual
t
n
d Algorithms for Federating Virtualized Infrastructures
Fig. 2. Role of the NOVI API
O
VI API receives requests from the NOVI GUI. The G
U
n
ce Editor - OIntEd [10], which was originally used to a
s
o
f the NOVI IM and subsequently was customized to a
l
u
ests for NOVI slices. In its current implementation, the N
O
a
g-and-drop interface for this instantiation process and all
o
t
ween instantiated objects. For example, a user can defi
n
o
ng with the characteristics for requested resources. For e
v
n
OWL document based on the NOVI IM which is sent t
o
H
TTP post request.
(
accessible online at http://novi-im.appspot.com/) the
u
a
ble ontologies in order to define the topology of the s
i
on Service(RIS)
Service (RIS) acts as a single point of contact for o
t
e
information about the status of virtual and subst
r
o
very encompasses locating and retrieving informa
t
a
lized substrate network in a decentralized way wit
R
IS uses the Request Handler to communicate with
e
rve resources and to obtain the resource advertisement
s
e
to query on the availability and the status of the resou
r
e
t information related to the access rights or the users.
s
of the NOVI information model to improve the preci
s
o
apply reasoning when selecting resources and service
e
d on semantic web technologies, namely Sesame [11],
t
a are stored in the Sesame database as RDF triples [
1
ool [13] for the conversion of triples to Java objects. T
h
the concepts in the NOVI IM are used also by the o
t
e
NOVI C&M architecture.
e
Mapping (IRM) Service
M
apping (IRM) service for NOVI will enable embed
d
t
opologies - resources (Virtual Networks - VNs) to
217
U
I is
s
sist
l
low
O
VI
o
ws
n
e a
v
ery
o
the
u
ser
lice
t
her
r
ate
t
ion
t
h a
the
s. It
r
ces
s
ion
s. It
for
1
2].
h
ese
t
her
d
ing
the
218 L. Lymberopoulos et al.
federated physical substrate network. This was initially formulated for a single
domain (infrastructure) as Virtual Network Embedding (VNE), an NP-Hard
combinatorial problem [14]. In the NOVI federated profile, VNE had to be extended
towards a multi-domain environment via graph spitting as in [15] and intelligent
selection of intra-domain mapping.
Evaluation and testing of the embedding procedure for NOVI experiments require
the appropriate representation of a VN request, formulated using the NOVI
Information Model. The IRM gathers information from the Resource Information
Service (RIS) and the Monitoring Service regarding available resources. As a first
step, user requests for VN resources are apportioned to infrastructures that are
members of a NOVI federation. Subsequently, single-domain VNE problems are
formulated, resulting into sub-optimal allocation of virtual resources within the
federated substrate.
A user may submit requests for standalone virtual resources, topologies of virtual
resources and specific services regarding virtual resources/topologies. These requests
may request specific mappings of virtual resources to substrate infrastructures. As
specified by the ProtoGENI RSpec [16]. VN requests may contain a complete (pre-
specified, bound), partial, or empty (free, unbound) mapping between virtual
resources and available physical (substrate) resources.
3.4 Policy Service
The Policy Service is used to provide the functionality of a policy-based management
system, where policies are used to define the behavior governing the managed
environment. As reported in [17], we plan to extend the Ponder2 policy framework
[18] with functions to support enforcement of mission policies. These will be used to
define the obligations of a member-infrastructure within a NOVI federation.
We currently provide support for (1) Access Control policies that specify what
rights users have on specific resources and (2) Event-Condition-Action policies
enforcing management actions upon events indicating failures or performance
degradation. Events are received by the Monitoring Service. Implementation details
are reported in NOVI Public Deliverable D2.2 [8].
3.5 Monitoring Service (MS)
One of the main challenges for a Monitoring Service (MS) in a heterogeneous
federated virtual environment is the diverse combination of monitoring tools deployed
within different infrastructures. To address this, NOVI developed generic Monitoring
Ontologies, enabling us to describe, parameterize and use diverse active and passive
monitoring tools provided within constituent federated infrastructures. Thus, users are
required to specify metrics to be measured and do not rely on monitoring tools.
MS collects information about specific resources and measurement metrics.
Monitoring can be performed on slivers (virtualized resources allocated to a user) or
on the physical substrate resources (hosts, links, paths, etc.). It is possible to obtain
passive monitoring information from resources or from repositories, and active
monitoring information as requested. Depending on usage scenario, MS can support
two main tasks: The first task is triggered by the Resource Information Service prior
NOVI Tools and Algorithms for Federating Virtualized Infrastructures 219
to resource allocation to collect monitoring and measurement information from the
substrate, which can be used by the IRM service to ensure that the constraints defined
in the resource requests are satisfied. The second task is used after the resource
reservation, to perform slice monitoring for diagnostic and watchdog purposes, i.e. to
check the current status of a given set of virtual resources across a NOVI federation.
MS supports three advanced high level monitoring tools, i.e. SONoMA [19],
HADES [20], and Packet Tracking [21]. These tools enable users to measure key
performance metrics of the network, for example the one-way delay, the round-trip
time, the packet loss, or the available bandwidth. Obviously, the MS can obtain from
hosts via command line SSH CPU utilization, memory consumption, disk usage etc.
In Fig. 3, we provide a screenshot of the MS GUI provided to users of NOVI, who
can choose from available metrics and specify required and optional parameters. Note
that users do not need to specify which monitoring tool will measure the selected
metric, as these can vary across infrastructures (testbeds) in a NOVI federation.
Measurements of selected metrics, associated with monitoring tasks, can be managed
individually, independently from the other monitoring task. The monitoring tasks can
be started, stopped or removed from the task list. The results of the measurements can
be read from the console of the GUI, or uploaded to a database within the Resource
Information Service, or trigger event-condition-action policies in the Policy Service.
Fig. 3. GUI of the NOVI Monitoring Service
3.6 Request Handler Service
The main purpose of this service is to perform two types of operations: (1) Handling
of resource allocation requests to the underlying platforms and (2) handling external
calls coming from testbeds that are members of a NOVI federation.
For the first type, the NOVI IM needs to be translated into the underlying platform
resource specification model. Given the key role played by SFA in the federation of
PlanetLab and FEDERICA for NOVI experiments, a translation needs to be
performed between NOVI IM concepts and the ones in SFA RSpec v2 [16].
Translation in the opposite direction is needed to handle remote calls from the
federated platforms. External calls from underlying platforms occur when the Resource
Information Service (RIS) needs updates with new information, i.e. the presence of new
220 L. Lymberopoulos et al.
resources or resource status notification updates received from the Monitoring Service
(MS). RIS will only store the static part of the information from the monitoring
ontology, while the dynamic parts will be obtained by directly calling MS.
The Request Handler, as shown in Fig. 1, communicates via RSpec with a server
running SFA. Since the SFA code was initially developed for PlanetLab, we just had
to installa private SFA server for the PlanetLab part of our testbed. However, there is
no SFA implementation for FEDERICA; thus we developed an appropriate
FEDERICA RSpec and an FEDERICA SFA Wrapper service acting as FEDERICA’s
Aggregate Manager (see Fig. 1). More information can be found in D2.2: First
Information and Data Models [8].
3.7 The NOVI Distributed Virtual Switch - NSwitch
The NSwitch distributed software complements NOVI’s federation architecture by
providing a unified way of interaction between heterogeneous domains at the data-
plane. It enables a virtual entity in one domain to be connected at protocol layer 2
(L2) with another virtual entity in a remote domain taking into account concurrence,
isolation, elasticity and programmability aspects.
The NSwitch was developed, deployed and tested over the PlanetLab – FEDERICA
testbed above. Its functionality was driven by the need to combine virtual resources
belonging to these two virtualization infrastructures that employ dissimilar
communication protocols and hypervisors. Notably, PlanetLab does not provide users
with data-plane connectivity options, using IP/BGP over the legacy Internet. By
contrast, FEDERICA provides users with data-plane network virtualization choices,
e.g. providing Juniper Logical Routers and Ethernet switches based on L2 VLAN
technology. User-configurable VLANs are carried by SDH 1 Gbps circuits
provisioned by NRENs and GÉANT into a controlled WAN environment, thus
enabling repeatability of experiments over the FEDERICA infrastructure.
In order to map PlanetLab slivers into an L2 broadcast domain we adopted an
approach similar to the one developed within the VINI [22] project in the US that
introduced a set of extensions to the PlanetLab kernel and tools. VINI used an
Ethernet over GRE – EGRE [23] mechanism to provide point-to-point virtual network
capabilities to user configured virtual resources over the Internet. NOVI’s NSwitch
enhanced VINI’s capabilities by introducing the Open vSwitch (OVS) [24] S/W in
PlanetLab’s host OS, thus enabling point-to-multipoint virtual links. OVS, compared
to the VINI multiple bridges, provides better management of multiple EGRE tunnels
within a host. Furthermore, distributed OVS instances can be centrally managed via
the OpenFlow protocol [25]. This feature will be adopted in NOVI’s Spiral 2 phase.
On the FEDERICA side, L2 data plane connectivity is provided by means of
VLANs used by Logical Routers, Switches and VMs. The NSwitch functionality
provides the mapping of EGRE key values of packets originating from PlanetLab
slivers to VLAN IDs.
3.8 Integration of NOVI C&M Services
In each platform (member of a NOVI federation) the C&M Services components of
the top layer in Fig. 1 interact with each other and communicate (1) northbound with
NOVI Tools and Algorithms for Federating Virtualized Infrastructures 221
the NOVI GUI and (2) south-bound with the middle layer (SFA). The north-bound
interface is the NOVI API, while the south-bound interface is the Request Handler
Service. Intra-domain C&M Services within the top layer exchange messages via an
Enterprise Service Bus - ESB [26]. Inter-domain C&M services can communicate (1)
via the Request Handler using SFA services (e.g. for slice creation across domains) or
(2) directly in a peer-to-peer mode via secure RPCs in cases that SFA mechanisms
were deemed as inadequate (e.g. for remote interactions of monitoring services).
An example of C&M service integration is the Slice Creation Use Case detailed in
NOVI Public Deliverable D4.2: Use Cases [27], which also provides an overview of
initial usage scenarios of the project. In summary, an authenticated experimenter is
authorized to use a set of resources across domains, as confirmed by the relevant per-
domain Policy Services. He may then request a desired topology using the NOVI GUI.
The virtual topology request is then passed to the IRM through the NOVI API. Prior
to solving the inter-domain VNE, IRM contacts RIS to identify available resources
that would fulfill the requirements imposed by the experimenter. RIS interacts with
the Monitoring Service to obtain information regarding the status (e.g. availability,
capacity, usage) of resources. Finally, when an appropriate mapping of virtual-to-
substrate resources is identified, reservation requests in the form of RSpecs are sent by
the Request Handler to the relevant testbed(s) slice manager(s).
NOVI developed a software integration framework for its C&M Services
architecture. It follows the Service Oriented Architecture complemented with the
Event Driven Architecture to enable synchronous and asynchronous communication
between components. The integration framework was implemented using Java
technologies. However it supports communication of components written in different
programming languages via a range or specific bridges such as: Jython [28], a Python
engine for Java; JRuby [29] for the Ruby language; JNI [30], a Java Native Interface
API for components written in C/C++.
4 NOVI Experimentation Testbed
To test and validate NOVI’s prototypes, a testbed environment was configured
consisting of private PlanetLab and FEDERICA resources. This testbed enables
NOVI software developers to run, test, refine and validate their software components
and prototypes, according to the experimentally driven methodology followed in the
project. NOVI developers are able to configure operational slices within the NOVI
testbed, in isolation from production services of the two virtualization platforms.
In fact, the testbed uses the actual FEDERICA substrate and virtualization services,
i.e. Juniper MX480 Logical Routers [31] and VMWare ESXi [32] Virtual Machines
(VMs). By contrast, the public PlanetLab could not be used as is for NOVI’s
experiments that require S/W upgrades, embedding custom code to C&M tools within
MyPLC [33] and root access rights to host hypervisors. Note that PlanetLab is a
widely used federated infrastructure [34], consisting of PLC (PlanetLab Central), PLE
(PlanetLab Europe) and PLJ (PlanetLab Japan), each with a single instance of
MyPLC. Experiments affecting PlanetLab’s host OS and C&M S/W are usually
performed on private testbed installations and this practice was also adopted in NOVI.
222 L. Lymberopoulos et al.
Fig. 4. Example of an experimentation slice in the NOVI testbed
Fig.4 presents the topology of one operational slice used to test control and
management plane components, detailed in NOVI Public Deliverable D4.1: Software
Architecture Strategy & Developers’ Guide [35]. This slice is comprised of three
FEDERICA core PoPs located in PSNC (Poznan, Poland), DFN (Erlangen, Germany)
and GARR (Milano, Italy). These are connected over the Internet via GRE tunnels to
private PlanetLab nodes in NTUA (Athens, Greece), ELTE (Budapest, Hungary) and
PSNC (Poznan Poland).
To isolate the slice in Fig. 4 from other NOVI slices using the same FEDERICA
core PoPs, Logical Routers are created on the Juniper MX480 routers. The open
source MyPLC (PlanetLab’s C&M software) is deployed at PSNC, managing the
private PlanetLab testbed.
An illustration of a typical slice deployed in the NOVI testbed is the NOVI-
MONITORING devoted for validating NOVI’s monitoring methods (active and
passive) and their corresponding tools. Measurements assembled via this slice are
depicted in Fig. 3.
5 Summary and Future Work
This paper reported a summary of NOVI’s current research outcomes: The ontology-
based NOVI information model, the advanced services within NOVI’s federated
Control & Management architecture and the distributed virtual switch architecture
(NSwitch). More technical details on the aforementioned work can be found in
NOVI's public deliverables and publications that are available at the project's website,
http://www.fp7-novi.eu.
NOVI’s research revealed a plethora of areas requiring further investigation. We
list below some of them:
Information Model: Need for constant updating of NOVI’s IM evolutionary
ontologies, e.g. incorporating NSwitch parameters. Short-term ontology
NOVI Tools and Algorithms for Federating Virtualized Infrastructures 223
enhancements based on feedback coming from validation of the integrated
prototype implementation.
GUI: Implementation of a user feedback mechanism and support for grouping of
graphical objects, thus simplifying the level of information details of user
requests.
Resource Information Service: Validation of the distributed architecture model
and support for more complex semantic queries that aim to provide efficient
resource discovery mechanisms, towards facilitation of virtual network
embedding processes.
Monitoring Service: Implementation of a mechanism allowing different
monitoring tools over dissimilar platforms to cooperate and contribute to multi-
domain measurements of the same metric.
Policy Service: Definition and deployment of role based access control (RBAC)
policies and enhancement of the policy engine to support enforcement of inter-
domain obligation Ponder2 policies
NSwitch: Integration of the NSwitch control plane with the other components of
NOVI's C&M plane.
We are currently at the end of Spiral 1 of the project, having deployed a first version
of an integrated prototype on the NOVI testbed (Section 4 of this paper). It is
expected that the Spiral 2 subsequent effort will complement functionality and
performance of NOVI's prototypes, based on Spiral 1 results obtained from validation
experiments on the NOVI experimental testbed.
Open Access. This article is distributed under the terms of the Creative Commons Attribution
Noncommercial License which permits any noncommercial use, distribution, and reproduction
in any medium, provided the original author(s) and source are credited.
References
[1] NOVI FP7 STREP Project, http://www.fp7-novi.eu
[2] Szegedi, P., Figuerola, S., Campanella, M., Maglaris, V., Cervelló-Pastor, C.: With
Evolution for Revolution: Managing FEDERICA for Future Internet Research. IEEE
Communications Magazine 47(7), 34–39 (2009)
[3] PlanetLab, http://www.planet-lab.org
[4] GÉANT, http://www.geant.net/pages/home.aspx
[5] Global Environment for Network Innovations (GENI), http://www.geni.net/
[6] D2.1: Information Models for Virtualized Architectures, http://www.fp7-novi.
eu/index.php/deliverables
[7] Web Ontology Language (OWL), http://www.w3.org/TR/owl-features
[8] D2.2: First Information and Data Models, http://www.fp7-novi.eu/index.
php/deliverables
[9] Slice Federation Architecture, v2.0, http://groups.geni.net/geni/
attachment/wiki/SliceFedArch
[10] Ontology instance editor - OIntEd, http://novi-im.appspot.com
[11] Openrdf Sesame, http://www.openrdf.org
[12] Resource Description Framework - RDF, http://www.w3.org/RDF/
[13] Alibaba, http://www.openrdf.org/alibaba.jsp
224 L. Lymberopoulos et al.
[14] Mosharaf Kabir Chowdhury, N.M., Boutaba, R.: Network Virtualization: State of the
Art & Research Challenges. IEEE Communications Magazine 47(7), 20–26 (2009)
[15] Houidi, I., Louati, W., Ameur, W.B., Zeghlache, D.: Virtual network provisioning
across multiple substrate networks. Elsevier Computer Networks 55, 1011–1023 (2011)
[16] RSpec, http://www.protogeni.net/trac/protogeni/wiki/RSpec2rac/
protogeni/wiki/RSpec
[17] Lymberopoulos, L., Grosso, P., Papagianni, C., Kalogeras, D., Androulidakis, G., van
der Ham, J., de Laat, C., Maglaris, V.: Managing Federations of Virtualized
Infrastructures: A Semantic-Aware Policy Based Approach. In: Proc. of 3rd IEEE/IFIP
International Workshop on Management of the Future Internet, Dublin, Ireland, May 27
(2011)
[18] Ponder2, http://ponder2.net
[19] Hullár, B., Laki, S., Stéger, J., Csabai, I., Vattay, G.: SONoMA: A Service Oriented
Network Measurement Architecture. In: Korakis, T., Li, H., Tran-Gia, P., Park, H.-S.
(eds.) TridentCom 2011. LNICST, vol. 90, pp. 27–42. Springer, Heidelberg (2012)
[20] https://wiki.man.poznan.pl/perfsonar-mdm/index.php/Hades_MA
[21] Santos, T., Henke, C., Schmoll, C., Zseby, T.: Multi-hop packet tracking for
experimental facilities. In: SIGCOMM 2010, New Delhi, India, August 30-September 3
(2010)
[22] VINI, http://www.vini-veritas.net
[23] Farinacci, D., Li, T., Hanks, S., Meyer, D., Traina, P.: RFC 2784, Generic Routing
Encapsulation (GRE) (March 2000)
[24] Open vSwitch, http://openvswitch.org/
[25] OpenFlow, http://www.openflow.org
[26] Chappell, D.: Enterprise Service Bus. O’Reilly (June 2004) ISBN 0-596-00675-6
[27] D4.2: Use Cases, http://www.fp7-novi.eu/index.php/deliverables
[28] Jython, http://www.jython.org
[29] JRuby, http://jruby.org
[30] Java Native Interface, http://java.sun.com/docs/books/jni/
[31] Juniper MX480, http://www.juniper.net/us/en/local/pdf/
brochures/1500027-en.pdf
[32] VMware ESXi, http://www.vmware.com/files/pdf/ESXi_
architecture.pdf
[33] MyPLC, http://www.planet-lab.org/doc/myplc
[34] PlanetLab federation, http://www.planet-lab.org/federation
[35] D4.1: Software Architecture Strategy and Developers’ Guide, http://www.fp7-
novi.eu/index.php/deliverables
... The high level overview of the NOVI Data, Control and Management architecture, reported in [8], is presented in Figure 1. In this section we will go through the lifecycle of a request and show how the different services of NOVI, contribute to a request, so as to explain the NOVI architecture. ...
Article
The NOVI Information Model (IM) and the corresponding data models are the glue between the software components in the NOVI Service Layer. The IM enables the communication among the various components of the NOVI Architecture and supports the various functionalities it offers. The NOVI IM consists of three main ontologies: resource, monitoring and policy ontology that have evolved over time to accommodate the emerging requirements of the NOVI architecture. This article presents the NOVI IM and its ontologies, together with an overview of how the NOVI software prototypes have benefited from using the IM.
... NOVI was set out to create a federation layer on top of existing network and computing FI platforms.Fig. 1 illustrates the NOVI federation approach, introduced in [22] . Utilizing NOVI's GUI, experimenters are able to request a slice of virtualized resources, spanning through the entire set of virtualized platforms that are members of the NOVI federation. ...
... NOVI's aim is to address the issue of vertical federation by designing and prototyping a service portfolio based on combined virtualized facilities from shared infrastructures at different layers. NOVI control framework[39]is being deployed in a federated testbed including the FEDERICA[8]infrastructure. The latter is a resource virtualization platform, augmented with network and computing facilities hosted in European NREN Points ofPresence.In this section the application and operation of the proposed resource mapping scheme over the FEDERICA experimental platform is demonstrated. The module responsible for resource mapping has been adapted to incorporate the embedding paradigm presented in the previous sections and enable allocation within the administrative domain of FEDERICA. ...
Article
Full-text available
Cloud computing builds upon advances on virtualization and distributed computing to support cost-efficient usage of computing resources, emphasizing on resource scalability and on demand services. Moving away from traditional data-center oriented models, distributed clouds extend over a loosely coupled federated substrate, offering enhanced communication and computational services to target end-users with Quality of Service (QoS) requirements, as dictated by the Future Internet vision. Towards facilitating the efficient realization of such networked computing environments, computing and networking resources need to be jointly treated and optimized. This requires delivery of user-driven sets of virtual resources, dynamically allocated to actual substrate resources within networked clouds, creating the need to revisit resource mapping algorithms and tailor them to a composite virtual resource mapping problem. In this paper, towards providing a unified resource allocation framework for networked clouds, we first formulate the optimal networked cloud mapping problem as a mixed integer programming (MIP) problem, indicating objectives related to cost-efficiency of the resource mapping procedure, while abiding by user requests for QoS-aware virtual resources. We subsequently propose a method for the efficient mapping of resource requests onto a shared substrate interconnecting various islands of computing resources, and adopt a heuristic methodology to address the problem. The efficiency of the proposed approach is illustrated in a simulation/emulation environment, that allows for a flexible, structured and comparative performance evaluation. We conclude by outlining a proof-of-concept realization of our proposed schema, mount over the European Future Internet test-bed FEDERICA, a resource virtualization platform augmented with network and computing facilities.
Conference Paper
This paper presents the cloud federation concept implemented by the NOVI project and ideas of possible extensions. The authors pay much attention to the networking aspect and available bandwidth provisioning systems which offer reliable network transfer services to the cloud systems and their federations.
Chapter
Monitoring and measurement is a fundamental building block for developing and testing new protocols, routing algorithms and networked applications. In a federated virtualized testbed they allow other service components and testbed-users to follow the current state of the network, and on the other hand they enable intelligent automatic decision-making, e.g. during the embedding of a virtual topology. However, it is not a trivial task to enable federated monitoring functionalities due to the cross-domain nature of the system. The heterogeneity of the federated networks (including network elements and monitoring tools) pose a major challenge. In this chapter we present a framework that tackles some of the most important related problems. We also introduce a specific ontology to describe monitoring and network measurement tasks. This semantic approach enables the flexible integration of a wide range of monitoring tools, metrics and databases. Our Monitoring Framework was created within the NOVI FP7 STREP project which federates two major virtualized testbeds, PlanetLab and Federica.
Article
Full-text available
Over the last two decades the importance of data networking for human beings and systems has increased beyond any expectation in size, complexity, and impact on society. Today, technology offers the ubiquitous and constant possibility of being connected to the Internet at a wide range of speeds. Traditional management solutions have up to now followed an evolutionary path, although the scale of the Internet and emerging novel architectures such as peer-to-peer, ad hoc networks, as well as virtualization-capable network infrastructures require focused and possibly revolutionary changes in management approaches. This article elaborates on challenges posed by the renaissance of virtualization as experienced in the planning, development, and operation of the FEDERICA infrastructure. The European Community cofunded project FEDERICA, like other worldwide initiatives such as FIND/GENI in the United States, NWGN in Japan, and the FIRE program in Europe, is supporting the development of the future Internet. FEDERICA extends the virtualization capabilities of the current hardware and software to provide a flexible infrastructure to host disruptive testing by networking researchers.
Article
Full-text available
Recently network virtualization has been pushed forward by its proponents as a long-term solution to the gradual ossification problem faced by the existing Internet and proposed to be an integral part of the next-generation networking paradigm. By allowing multiple heterogeneous network architectures to cohabit on a shared physical substrate, network virtualization provides flexibility, promotes diversity, and promises security and increased manageability. However, many technical issues stand in the way of its successful realization. This article investigates the past and the state of the art in network virtualization along with the future challenges that must be addressed to realize a viable network virtualization environment.
Conference Paper
Full-text available
This paper presents our work toward organizing and managing various forms of federations of virtualized infrastructures. We adopt the Ponder2 policy framework and the SMC architecture as a powerful engineering approach, which we apply to semantic-aware management of federations of Future Internet (FI) virtualized infrastructures. To cater for context-awareness, we plan for a common information model, based on the Network Description Language (NDL), capturing a common set of abstractions of virtualized resources and services, nodes, routers and switches, custom network topologies with specific bandwidth demands, etc. To handle management of generic complex federated environments, we employ structural patterns to model federations as graphs, whose vertices represent SMCs and edges denote the type of relationship between them. We give an illustration of such structures corresponding to existing FI experimental platforms in the US and Europe and we provide examples containing inter-domain management responsibilities as missions. Finally, we propose to augment the Ponder2 framework with single & multi-domain resource provisioning capabilities, enabling efficient sharing of virtualized networked facilities among federation users.
Conference Paper
Full-text available
The Internet has become a complex system with increasing numbers of end-systems, applications, protocols and types of networks. Although we have a good understanding of how data is transferred over the network we cannot observe what happens with our data after sending and before receiving it - how packets traverse through the network and with which QoS characteristics remains unknown. Towards this objective we have developed a multi-hop packet tracking system intended to be used in experimental facilities, such as PlanetLab, where we have made our first tests. This paper describes our packet tracking realization and the results from our prototype implementation.
Article
Full-text available
The Internet has become a complex system with increasing numbers of end-systems, applications, protocols and types of networks. Although we have a good understanding of how data is transferred over the network we cannot observe what happens with our data after sending and before receiving it - how packets traverse through the network and with which QoS characteristics remains unknown. Towards this objective we have developed a multi-hop packet tracking system intended to be used in experimental facilities, such as PlanetLab, where we have made our first tests. This paper describes our packet tracking realization and the results from our prototype implementation.
Conference Paper
Full-text available
To characterize the structure, dynamics and operational state of the Internet it requires distributed measurements. Although in the last decades several systems capable to do this have been created, the easy access of these infrastructures and orchestration of complex measurements is not solved. We propose a system architecture that combines the flexibility of mature network measurement infrastructures such as PlanetLab or ETOMIC with the general accessibility and popularity of public services like Web based bandwidth measurement or traceroute servers. To realize these requirements we developed a multi-layer architecture based on Web Services and the basic principles of SOA, which is a very popular paradigm in distributed business application development. Our approach opens the door to perform complex network measurements, handles heterogeneous measurement devices, automatically stores the results in a public database and protects against malicious users as well. To demonstrate our concept we developed a public prototype system, called SONoMA. Comment: Technical Report
Article
> Map of our Site 1999-02-01T00:00Z may be abbreviated as 7. Acknowledgements This specification is the work of the W3C RDF Model and Syntax Working Group. This Working Group has been most ably chaired by Eric Miller of the Online Computer Library Center and Bob Schloss of IBM. We thank Eric and Bob for their tireless efforts in keeping the group on track and we especially thank OCLC and IBM for supporting them and us in this endeavor...