![](bg9.png)
configuration information for dodai regularly at site-
A and restored them after reconstruction of the
OpenStack at site-B using dodai (Figure 15).
Figure 15: Cloud Migration.
Moreover, we could migrate any software
supported by dodai-deploy, i.e., OpenStack, Hadoop,
GridEngine and Eucalyptus.
2) Performance of inter-cloud object storage
PUT and GET
By making the object storage service OpenStack
swift network-aware, the inter-cloud object storage
is almost equal in performance to local object
storage for PUT and GET, which is described in
(Yokoyama et al., 2012b).
5.5 User Scenario According to U4
(Service Continuity)
Normally, if a provider suspends its business, its
customers need to re-register with different
providers for similar services. To avoid such a
situation, resources, applications, and customer ID
data for the services provided by one provider can be
transferred to the cloud systems of other providers in
advance. Then, if its business is suspended, its
consumers can use similar services provided by the
other providers.
1) Cloud migration
As described in the previous section, however it
not necessary for dodai to regularly store a user
database, snapshots, or configuration information.
2) Performance of inter-cloud storage PUT and
GET performance
Same as in the previous section.
6 CONCLUSIONS
We proposed a solution called cloud on demand and
described a prototype implementation based on the
dodai and colony projects. The cloud-on demand
was proved to be feasible in the actual user scenarios
in one data center. This architecture can be extended
to wide area networks using SINET L2VPN and
VPLS services if we plug the upper link from the
OpenFlow switches into the SINET directly.
We are now constructing a new prototype of
cloud on demand upon SINET, and we will evaluate
its performance in this wide area network
environment.
ACKNOWLEDGEMENTS
We would like to thank all the contributors to the
dodai project, especially Shin-ichi Honiden,
Yoshitaka Kuwata, Masashi Ochida, Osamu
Habuka, Takahiro Shida, Guan Xiaohua, Motonobu
Ichimura, Takahiko Yuzawa and Daishi Kimura.
REFERENCES
GICTF: http://www.gictf.jp/index_e.html.
Hiroshi Sakai, "Standardization Activities for Cloud
Computing", NTT Technical review, Vol. 9 No. 6,
June 2011.
Shigetoshi Yokoyama, Nobukazu Yoshioka, “Cluster as a
Service for self-deployable cloud applications”,
pp.703-704, Cluster, Cloud and Grid Computing
(CCGrid), 2012 12th IEEE/ACM International
Symposium, 2012a.
Shigetoshi Yokoyama, Nobukazu Yoshioka, "An
Academic Community Cloud Architecture for Science
Applications," pp. 108-112, 2012 IEEE/IPSJ 12th
International Symposium on Applications and the
Internet (SAINT), 2012b.
Shigetoshi Yokoyama, Nobukazu Yoshioka, Motonobu
Ichimura, “Intercloud Object Storage Service: Colony”,
pp. 95-98, CLOUD COMPUTING 2012, The Third
International Conference on Cloud Computing, GRIDs,
and Virtualization, 2012a.
SINET: http://www.sinet.ad.jp/
index_en.html?lang=english.
Dodai-compute: https://github.com/nii-cloud/dodai-
compute.
Dodai-deploy: https://github.com/nii-cloud/dodai-deploy.
Puppet: http://puppetlabs.com/
Mcollective: http://puppetlabs.com/mcollective/
Colony: https://github.com/nii-cloud/colony.
OpenStack: http://openstack.org/
Shigetoshi Yokoyama, Nobukazu Yoshioka and
Motonobu Ichimura, "A Network-aware Object
Storage Service", pp.556-561, The 2nd International
Workshop on Network-aware Data Management to be
held in conjunction with SC12, 2012b.
On-demandCloudArchitectureforAcademicCommunityCloud-AnotherApproachtoInter-cloudCollaboration
669