Interoperability Standards for Cloud Architecture

Claus Pahl, Li Zhang and Frank Fowley

IC4, Dublin City University, Dublin, Ireland

Keywords:

Cloud Computing, Cloud Architecture, Cloud Interoperability, Cloud Standards.

Abstract:

Enabling cloud infrastructures to evolve into a transparent platform raises interoperability issues. Interoper-

ability requires standard data models and communication technologies compatible with the existing Internet

infrastructure. To reduce vendor lock-in situations, cloud computing must implement common strategies re-

garding standards, interoperability and portability. Open standards are of critical importance and need to be

embedded into interoperability solutions. Interoperability is determined at the data level as well as the service

level. Relevant modelling standards and integration solutions shall be analysed in the context of clouds.

1 INTRODUCTION

Enabling clouds to be merged and connected cre-

ates interoperability problems (Armbrust et al., 2009;

Buyya et al., 2011). Interoperability requires standard

data models and communication technologies com-

patible with existing Internet infrastructure. Public

cloud services are available to the public. Hybrid

clouds are an integrated cloud services arrangement

that includes provision of compute resources from

more than one source. Hybrid architectural models

may be vertically partitioned (e.g. data stored pri-

vately) or horizontally partitioned (e.g. using public

cloud to prototype a new device view of a service

in parallel with an existing implementation). Inter-

operability solutions for ﬂexible composition, migra-

tion and portability are sought (451 Group, 2010). To

reduce vendor lock-in, cloud computing must imple-

ment universal strategies regarding standards, inter-

operability and portability. Open standards are of crit-

ical importance and need to be embedded into inter-

operability solutions (DMTF, 2010). Interoperability

is determined at the data level as well as the service

level (OMG, 2009; Cloud Standards, 2013). The ob-

jectives here include the review of relevant standards

for cloud architecture interoperability and analysing

the overall maturity of the technology and determin-

ing current trends and shortfalls.

Typically, cloud computing consists of the three

cloud layers infrastructure (IaaS), platform (PaaS)

and software (SaaS) as service (Armbrust et al., 2009;

Buyya et al., 2011). In addition, we can differen-

tiate between (hardware or software) resources pro-

vided in a traditional way, and the . . . as a Service ver-

sion of them, which considers virtualization, multi-

tenancy and elasticity as the main properties (Wang

et al., 2009). Different usage models can be derived

from the combinations of the layers. For instance, an

application over a platform: a software provider can

develop an application (e.g., a content management or

delivery system) and publish it. That allows it to be

deployed using some platform software (Web appli-

cation container or database) and offered as a service

(PaaS). The different usage models rely on interop-

erability solutions. Service-based abstractions (APIs)

need to be aligned with common (Web) service de-

scription, modelling and composition standards.

Interoperability concerns arise in different situa-

tions: Firstly, interoperability between layers - stan-

dardised APIs are necessary to allow higher cloud

layers to link to a range of services provided at the

lower layers, e.g. platform implementations to uni-

formly link to IaaS offerings. Roles of importance

are service provider and service user. Secondly, in-

teroperability within layers - standards are required

to allow components in a layer to interact and be ex-

changeable. Non-service interactions need to be sup-

ported, e.g. where a software developer combines dif-

ferent platforms in the development of a new system.

2 INTEROPERABILITY

STANDARDS

Standards are necessary to consolidate efforts in a

technology domain and to enable interoperability in

123

Pahl C., Zhang L. and Fowley F..

Interoperability Standards for Cloud Architecture.

DOI: 10.5220/0004366901230126

In Proceedings of the 3rd International Conference on Cloud Computing and Services Science (CLOSER-2013), pages 123-126

ISBN: 978-989-8565-52-5

 2013 SCITEPRESS (Science and Technology Publications, Lda.)

any technology domain. An overview and a categori-

sation of standards relevant to interoperability in the

cloud computing context that we cover here is:

• Service modelling: Open-SCA (service compo-

sition and interaction), USDL/SoaML/CloudML

(multi-view services), EMML (mashups)

• Service interfaces: OCCI (infrastructure manage-

ment), CIMI (infrastructure management), EC2

(de-facto standard), TOSCA (portability), CDMI

(data management)

• Infrastructure: OVF (virtual machines)

For each standard, we provide background about ori-

gins, support and purpose, the intended usage, and an

analysis of the relevance for interoperability consid-

erations. Providing a comprehensive overview of all

standards is not the objective. We have singled out

those that represent speciﬁc aspects well. Well-known

services standards and security standards shall not be

covered due to space reasons.

The OASIS Open Composite Services Architec-

ture (CSA) aims to simplify SOA application devel-

opment (Open CSA, 2012) connected to the Service

Component Architecture (SCA) and Service Data Ob-

jects (SDO) families of speciﬁcations. The SCA As-

sembly Model is a framework to describe service co-

ordination and interaction that ties in service compo-

sition with common software architecture concerns.

The CSA standards can be utilised as is or can serve

as input for any composition and assembly language.

The speciﬁcations on the SCA Assembly Model are

very relevant for interoperability. Data interoperabil-

ity is an equally important concern in Open-CSA.

Data could be represented in Service Data Objects.

Enterprise Mashups combine and remix data from

databases, spreadsheets, websites, Web Services,

RSS/Atom feeds, and unstructured sources that de-

liver actionable information. The Open Mashup Al-

liance (OMA) is in charge of the Enterprise Mashup

Markup Language (EMML) (EMML, 2012). It can

support enterprise mashup implementations, improve

mashup portability of mashup designs, and increase

the interoperability of mashup solutions. EMML is

designed to be complimentary to and integrated with

languages like JavaScript, Java, Groovy, and Ruby

via scripting. It is particularly suited for interop-

erability issues related to mashup creation. It sup-

ports lightweight and integrative service assembly

and, therefore, represents speciﬁc modelling and in-

tegration concerns.

The aim of the Uniﬁed Service Description Lan-

guage (USDL) is describing general parts of techni-

cal and business services to allow services to become

tradable and consumable (USDL, 2012). Technical

services are considered services based on WSDL,

REST or other technical speciﬁcations. Business ser-

vices are deﬁned as business activities that are pro-

vided by a service provider to a service consumer to

create value for the consumer. The USDL deﬁnition

aims at complementing the technical language stack

by adding required business and operational informa-

tion. The targeted cloud stakeholders for USDL are

service providers, infrastructure providers, service as-

semblers and service consumers. USDL deﬁnes an

interoperability-centriclanguage that enables its users

to model arbitrary services and to integrate with ex-

isting standards. The aim to address service mod-

elling and support mappings to different standards

makes this interesting for interoperable cloud mod-

elling. This enables new business models for service

brokerage because services can automatically be of-

fered, delivered, executed, and composed from ser-

vices of different providers.

Another development to be considered in this

context includes SoaML (standardised by OMG, see

http://www.omg.org/spec/SoaML/), which falls into

the same category as USDL in our categorisation as

a service description and modelling language. While

still under development (and thus far from being stan-

dardised), CloudML (http://www.cloudml.org/) is a

language more speciﬁc to clouds.

OCCI (infrastructure lifecycle management) is

now the ﬁrst of four cloud-speciﬁc standards, also in-

cluding CIMI (like OCCI on infrastructure manage-

ment), TOSCA (portability and cloud-bursting), and

CDMI (data management), interleaved with a note

on EC2 as a proprietary solution that has become a

de-facto standard. Providers offer IaaS solutions to

enhance elastic capacity, where server instances are

executed in their proprietary infrastructure and billed

on a utility computing basis. For the infrastructure

layer this means that typically virtual machines on a

per-instance per-hour basis are the units. The OGF

OCCI working group provides an API speciﬁcation

for the management of cloud computing infrastruc-

ture to capture these concerns (OCCI, 2012). The

scope is a comprehensive range of high-level func-

tionality required for the life-cycle management of

virtual machines (or workloads) running on virtual-

ization technologies (such as containers) supporting

service elasticity. OCCI provides an API for inter-

facing IaaS cloud computing facilities, which is suf-

ﬁciently complete to facilitate the implementation of

interoperable implementations. While targeting IaaS

concerns, it can foster interoperability endeavours at

higher levels. The scope of OCCI is high-level func-

tionality required for lifecycle management. This is

in part realised through coverage of existing propri-

CLOSER2013-3rdInternationalConferenceonCloudComputingandServicesScience

124

etary APIs. While the focus is on the upper cloud

stack layers for the section presented in Figure 1, it is

nonetheless a suitable framework for interoperability

concerns at the interface of infrastructure services.

Similar to OCCI, the CIMI - Cloud Infrastruc-

ture Management Interface from DMTF addresses in-

frastructure management. CIMI which addresses the

runtime maintenance and provisioning of cloud ser-

vices. The scope of the CIMI standard covers core

IaaS functionality, addressing deploying and manag-

ing virtual machines and other artifacts such as vol-

umes, networks, or monitoring. Once interfaced to

the IaaS provider, the information that needs to be

processed to manage a cloud service can be discov-

ered iteratively, including the metadata describing ca-

pabilities and resource constraints. Most develop-

ers use with the CIMI REST/HTTP-based protocol,

the current interface binding to the model (others are

expected later). This delivers standard HTTP status

codes and supports JSON and XML serialization for-

mats. CIMI, if widely used, would allow organisa-

tions to design cloud-based business solutions being

assured that management (and governance) processes

will not be compromised if the business solution is

moved to another (standards-based) IaaS provider.

While OCCI and CIMI are similar standards,

Amazon EC2 as a proprietary solution needs to be

considered as well. Amazon Elastic Compute Cloud

(Amazon EC2) provides resizeable computing capac-

ity servers in Amazon’s data centers to build and host

software systems. They allow access to components

and features using a web-based GUI, command line

tools and APIs . EC2 comes with a rich ecosystem.

Some open-source standards are pushed by Amazon

competitors to regain market shares.

Supported by OASIS, the TOSCA framework

aims to enhance the portability of cloud applications

and services. TOSCA enables interoperable descrip-

tion of application and infrastructure cloud services,

the relationships between parts of the service, and the

operational behaviour of these services (such as de-

ploy, patch, shutdown) independent of the supplier

creating the service, and any particular cloud provider

or hosting technology. TOSCA also aims to sup-

port higher-level operational behaviour to be associ-

ated with cloud infrastructure management. The core

concept behind TOSCA is cloud bursting, which is

the ability to move workloads between public and pri-

vate cloud infrastructures in a transparent way. There

seems to be some discussion about the prospect of

success, with large IaaS providers not having joined

the consortium yet. The core to the solution would be

a hypervisor-agnostic portability mechanism, which

requires IaaS compliance. TOSCA also needs to be

observed as a vendor initiative in the context of open-

source activities like OpenStack gaining momentum.

The CDMI, the Cloud Data Management Inter-

face by SNAI targets cloud storage. CDMI is a stan-

dard for self-provisioning, administering and access-

ing cloud storage. CDMI deﬁnes RESTful HTTP op-

erations for accessing the capabilities of the Cloud

storage system. CDMI deﬁnes the functional inter-

face that applications use to create, retrieve, update

and delete data elements from the cloud. As part of

this interface, a client can discover the capabilities

of the cloud storage offering and use this interface

to manage containers and the data that is placed in

them. In addition, metadata can be set on containers

and their contained data elements through this inter-

face. Compared to OCCI and OVF, CDMI speciﬁ-

cally targets data migration and format immigration.

Although CDMI can also be used for task manage-

ment, this would need extensive rules to be deﬁned.

OVF is the most critical and successful standard

speciﬁc to cloud infrastructure. The Open Virtu-

alization Format (OVF) describes an open, secure,

portable, efﬁcient, and ﬂexible format for the packag-

ing and distribution of one or more virtual machines

(OVF, 2012). OVF features include optimized distri-

bution and portability of virtual appliances. It sup-

ports compression for more efﬁcient package trans-

fers, content veriﬁcation and integrity checking, and

provides a basic scheme for the management of soft-

ware licensing. It supports vendor and platform inde-

pendence as it does not rely on the use of a speciﬁc

host platform, virtualization platform, or host/guest

operating system. It is also designed to be extended

as the industry moves forward with virtual appliance

technology. The OVF format provides a complete

speciﬁcation of a virtual machine, which may include

multiple virtual disks.OVF is a portable format that

allows users to deploy VMs in any hypervisor that

supports OVF. As such, it sits at the core of resource

management in the infrastructure provisioning layer,

overcoming previous deﬁciencies in standardised so-

lutions such as VMDK. For supporting interoperabil-

ity as a standard for this technical context, other fea-

tures like localisation are important.

3 DISCUSSION AND

CONCLUSIONS

Interoperability between clouds is vital for the fur-

ther development of the cloud ecosystem and mar-

ket. While standards for the Web Services context are

abundant, more speciﬁc standards for the cloud com-

puting domain reﬂect the current maturity.

InteroperabilityStandardsforCloudArchitecture

125

• A number of standards for the lower infrastructure

layers apply to respective cloud computing tech-

nologies. They address interoperability solutions

for speciﬁc aspects like virtual machine manage-

ment or data management.

• For platforms and services, the respective (Web)

service standards are still of relevance. Standards

exist, beyond the core Web services platform, that

can further the development of platform and soft-

ware services from existing offers. Generic ser-

vice solution can provide a starting point where

cloud-speciﬁc standards are lacking.

In addition, it is worth looking at a number of differ-

ent concerns that help us to judge the state of stan-

dardisation and it’s impact on interoperability: or-

ganisations behind standards and their domain, stake-

holders involvedthrough standards, and standards and

open-source/proprietary solutions

Firstly, by looking at the organisations behind the

standards, we can also observe that while the Web

services domain is primarily dominated by W3C and

OASIS in terms of standardisation, the situation in

cloud computing is more diverse. Some of the or-

ganisations active include DMTF (management of

distributed IT systems), the OGF (grid computing),

the OMG (middleware), SNIA (storage), OASIS (ser-

vices), OCC (cloud), as well as national (e.g. NIST)

and sector-speciﬁc (e.g. ETSI - telecoms) organisa-

tions. Currently, there is a dominance of infrastruc-

ture and lower-level management. Secondly, stake-

holders are yet another perspective that we can look

at. We have referred to stakeholders in the review and

discussion of standards where relevant to differenti-

ate the different interoperability needs of stakehold-

ers in clouds as multi-organisational, multi-role en-

vironments. While the infrastructure standards target

clearly software developers, the more generic service-

oriented standards are more at the interface (as-a-

service) level, targeting service providers and con-

sumers. Particularly combined roles, such as pro-

sumers or aggregators that are providers and con-

sumers beneﬁt from the recent service description

and modelling standards. Thirdly, while standards

can achieve interoperability, often de-facto standards

emerge from open-source or proprietary solutions.

We dicussed OCCI and CIMI as standards in a context

where OpenStack is a compliant open-source frame-

work, all competing with Amazon EC2.

By looking at the standards we reviewed here for

indications of future standardisation needs, emerging

from the categorisation of standards are the following

observations: (i) Modelling under incorporation of a

variety of standards can support migration and, con-

sequently, the uptake of cloud computing solutions.

(ii) Composition, e.g. through mashups, is becoming

of increasing importance to provide a market for ba-

sic and composite offering where providers and ag-

gregators compete. (iii) Quality of Service and Ser-

vice Level Agreement standardisations beyond secu-

rity concerns in the cloud are actually largely lacking.

REFERENCES

451 Group (2010). Cloud Computing ’As-a-service’ market

sizing. Report II

Armbrust, M., Fox, A., Grifﬁth, R., Joseph, A.D., Katz, R.,

Konwinski, A., Lee, G., Patterson, D., Rabkin, A.,

Stoica, I. and Zaharia, M. (2009). A view of cloud

computing. Comm. ACM 53(4), 50-58.

Buyya, R., Broberg, J., and Goscinski, A.(2011). Cloud

Computing - Principles and Paradigms. Wiley.

CloudCom (2011). Market Implementation of Cloud In-

teroperability and Portability Research in IaaS and

PaaS. Retrieved from: http://www.cloud4soa.eu/

workshop2011

Cloud Standards (2013). Retrieved from: http://

cloud-standards.org/wiki/index.php?title=Main Page

Distributed Management Task Force (2010). Interoperable

Clouds. Retrieved from: http://www.dmtf.org/sites/

default/ﬁles/standards/documents/ DSP-IS0101 1. 0.

0.pdf

Enterprise Mashup Markup Language (2012). Retrieved

from: http://www.openmashup.org/omadocs/v1.0/

index.html

OCCI Open Cloud Computing Interface (2012). Retrieved

from: http://occi-wg.org/

OMG Object Management Group (2009). Cloud Interop-

erability Roadmaps Session. Retrieved from: http://

www.omg.org/ news/ meetings/ tc/ ca/ special-events/

Cloud Interop Roadmaps.htm

Open CSA - Open Composite Services Architecture (2012).

Retrieved from: http://www.oasis-opencsa.org/

OVF Open Virtualization Format (2012.) Retrieved from:

http:// www.dmtf.org/ sites/ default/ ﬁles/ standards/

documents/DSP0243 1.1.0.pdf

Pahl, C. (2005) Layered ontological modelling for web

service-oriented model-driven architecture. European

Conference on Model Driven Architecture Founda-

tions and Applications ECMDA05, 88-102.

Pahl, C., Giesecke, S. and Hasselbring, W. (2009). An

ontology-based approach for modelling architectural

styles. European Conference on Software Architecture

ECSA’2007, 60-75.

Uniﬁed Service Description Language (2012.) Retrieved

from: http://www.w3.org/2005/Incubator/usdl/

Wang, M.X., Bandara, K.Y. and Pahl, C. (2009). Inte-

grated constraint violation handling for dynamic ser-

vice composition. IEEE International Conference on

Services Computing SCC’09.

CLOSER2013-3rdInternationalConferenceonCloudComputingandServicesScience

126