logo epfl
 Nicolas Bonvin
home page
 English only
     EPFL > I&C > LSIR > Nicolas Bonvin > Scarce
 sommaire

Contact

Projects

Publications

Teaching

Short biography

Blog

Twitter logo

Service-oriented architecture (SOA) paradigm for orchestrating large-scale distributed applications offers significant cost savings by reusing existing services. However, the high irregularity of client requests and the distributed nature of the approach may deteriorate service response time and availability. Static replication of components in datacenters for accommodating load spikes requires proper resource planning and underutilizes the cloud infrastructure. Moreover, no service availability guarantees are offered in case of datacenter failures. In this paper, we propose a cost-efficient approach for dynamic and geographically-diverse replication of components in a cloud computing infrastructure that effectively adapts to load variations and offers service availability guarantees. In our virtual economy, components rent server resources and replicate, migrate or delete themselves according to self-optimizing strategies. We experimentally prove that such an approach outperforms in response time even full replication of the components in all servers, while offering service availability guarantees under failures.

Cloud computing is deemed to replace high capital expenses for infrastructure with lower operational ones for renting cloud resources on demand by the application providers. However, with static resource allocation, a cluster system would be likely to leave 50% of the hardware resources (i.e. CPU, memory, disk) idle, thus baring unnecessary operational expenses without any profit (i.e. negative value flows). Moreover, as clouds scale up, hardware failures of any type are unavoidable.
A successful online application should be able to handle traffic spikes and flash crowds efficiently. Moreover, the service provided by the application needs to be resilient to all kinds of failures (e.g. software stales, hardware, rack or even datacenter failures, etc.). A naive solution against load variations would be static over-provisioning of resources, which would result into resource underutilization for most of the time. Resource redundancy should be employed to increase service reliability and availability, yet in a cost-effective way. Most importantly, as the size of the cloud increases its administrative overhead becomes unmanageable. The cloud resources for an application should be self-managed and adaptive to load variations or failures.
We propose a middleware ("Scattered Autonomic Resources", referred to as Scarce) for supple sharing to avoid stranded and underutilized computational resources that dynamically adapts to changing conditions, such as failures or load variations. Our middleware simplifies the development of online applications composed by multiple independent components (e.g. web services) following the Service Oriented Architecture (SOA) principles. We consider a virtual economy, where components are treated as individually rational entities that rent computational resources from servers, and migrate, replicate or exit according to their economic fitness. This fitness expresses the difference between the utility offered by a specific application component and the cost for retaining it in the cloud.
The server rent price is an increasing function of the utilization of server resources. Moreover, components of a certain application are dynamically replicated to geographically-diverse servers according to the availability requirements of the application.
Our approach combines the following unique characteristics:

  • Adaptive component replication for accommodating load variations.
  • Geographically-diverse placement of clone component instances.
  • Cost-effective placement of service components for supple load balancing.
  • Decentralized self-management of the cloud resources for the application.
Having implemented a full prototype of our approach, we experimentally prove that it effectively accommodates load spikes, it provides a dynamic geographical replica placement without thrashing and cost-effectively utilizes the cloud resources. Specifically, we found that our approach offers lower response time even than full replication of the service components to all servers.

 

Publications

  Autonomic SLA-driven Provisioning for Cloud Applications
Nicolas Bonvin, Thanasis G. Papaioannou, Karl Aberer
In the Proc. of the 11th IEEE/ACM International Symposium on Cluster, Cloud and Grid Computing 2011 (CCGRID2011), May 23-26, 2011, Newport Beach, CA, USA.
  An economic approach for scalable and highly-available distributed applications
Nicolas Bonvin, Thanasis G. Papaioannou, Karl Aberer
the Proc. of the 3rd IEEE International Conference on Cloud Computing 2010 (CLOUD2010), July 5-10, 2010, Miami, FL, USA.

© 2007-2011 Nicolas Bonvin | Last Modified 2011-05-30 14:59:04