GENI-related research activities of CSE, Aalto Zhonghong Ou Post-doc researcher Department of Computer Science and Engineering (CSE) Aalto University, Finland
Internet of Things Mario Di Francesco
GENI Meeting at KTH Mario Di Francesco ( ) – September 15, 2014 Heterogeneous and multimedia data in the Internet of Things How to store data of different types or even multimedia? What is the impact on performance?
GENI Meeting at KTH Mario Di Francesco ( ) – September 15, 2014 A storage infrastructure for heterogeneous and multimedia data General data model Document-oriented database infrastructure –support for replication –live updates –web-friendly querying system –support for binary data –metadata Mario Di Francesco, Mayank Raj, Na Li, and Sajal K. Das, "A Storage Infrastructure for Heterogeneous and Multimedia Data in the Internet of Things", The 2012 IEEE International Conference on Internet of Things (iThings 2012), Pages 26-33, November 2012 { "ts": " T13:31:37.459Z", "tags": ["POSTPROCESS"], "device_id": "OO14.4FO1.OOOO.O1AB", "sensor_id": [1,2], "data": [700,15.3], "_attachments": { "hello.txt”: { "content_type": "text/plain", "data": "SGVsbG8gd29ybGQh" }
GENI Meeting at KTH Mario Di Francesco ( ) – September 15, 2014 Database performance for IoT data What is the best solution to store IoT data in the cloud? –performance evaluation of different classes of databases Thi Anh MaiPhan, Jukka K. Nurminen, and Mario Di Francesco, “Cloud Databases for Internet-of-Things Data”, The 2014 IEEE International Conference on Internet of Things (iThings 2014), September 2014 Bulk insert latencyMultimedia insert and query latency
CIVIS Jukka K. Nurminen
Delay-sensitive mobile cloud Dedicated radio DSRC Cellular technology LTE Key idea: No new radios Processing in the cloud Short delay vs. resource use Topics: Optimal way to update cloud data Distributed cloud Business: Nokia/HERE cloud SMEs for new apps
CIVIS Social network and big data analysis for sustainable energy use Transaction to a distributed energy paradigm. Empowerment of local communities. ICT as enablers of sustainable social dynamics. Social dimension relevant to obtain CO 2 emissions reduction, energy efficiency and to achieve social goals. Energy System
Mobile Cloud Gaming Matti Siekkinen
(Mobile) Cloud Gaming Game rendered in the cloud and streamed to an end-user device through a thin client Latency is a key challenge: even 100 ms can be too much for the most demanding games Extremely distributed cloud infrastructure proven to be beneficial using a prototype in test scenarios -Eg. Cloudlets over Wi-Fi, or LTE with server in operator premises TODO: scalability and overall plausibility tests would require access to a real-world test network such as the GENI -How sparse/dense would the cloud network need to be to support even the most demanding games?
SIGMONA (SDN Concept in Generalized Mobile Network Architectures) Sakari Luukkainen
SIGMONA Cloud computing has been emerging as a promising approach to reduce cost for mobile operators Cloudification of the mobile network has momentums One significant source of expense is the use of dedicated network hardware to provide the required services Solution: Network Function Virtulisation (NFV) Focus -Distribution of cloud elements in the architecture of a mobile network -VM migration and its requirements and performance between cells or regions
Performance evaluation of public clouds Zhonghong Ou
Performance valuation of public clouds Amazon EC2 & Rackspace Cloud Introduced in 2006 Provisioning various categories of instances, diversified types of instances within the same category Hardware heterogeneity likely from Hardware upgrade and replacement Research problems Homogeneous vs. heterogeneous? Performance variation? Did experiments in Amazon EC2 and Rackspace for two periods 2011 & 2012
Findings Amazon EC2 uses diversified hardware to host the same type of instances. Hardware diversity is the primary culprit for performance variation in the cloud. Different VM scheduling mechanisms are used in EC2, which exacerbates performance variations, especially for networking related operations. In general, the variation between the fast instances and slow instances can reach 40%. In some applications, the variation can even approach up to 60%. By selecting fast instances within the same instance type, Amazon EC2 users can acquire up to 30% of cost saving, if the fast instances have a relatively low probability.
Related publications [1] Z. Ou, H. Zhuang, J. K. Nurminen, A. Ylä-Jääski, and P. Hui. Exploiting hardware heterogeneity within the same instance type of Amazon EC2. 4th USENIX Workshop on Hot Topics in Cloud Computing (HotCloud), (covered by BBC News, The Register, ACM TechNews etc) [2] Z. Ou, H. Zhuang, A. Lukyanenko, J.K. Nurminen, P. Hui, V. Mazalov, A. Yla-Jaaski. Is the Same Instance Type Created Equal? Exploiting Heterogeneity of Public Clouds," IEEE Transactions on Cloud Computing, vol.1, no.2, pp , July-December [3] H. Zhuang, X. Liu, Z. Ou, and K. Aberer. Impact of instance seeking strategies on resource allocation in cloud data centers IEEE Sixth International Conference on Cloud Computing (CLOUD ’13), 27 – 34, June