Towards autonomic computing: service discovery and web hotspot rescue
Henning G. Schulzrinne, Weibin Zhao · 2006
Autonomic computing is a vision that addresses the growing complexity of computing systems by enabling them to manage themselves without direct human intervention. This thesis studies two related problems, service discovery and web hotspot rescue, which can serve as a building block and a prototype for autonomic networking and distributed systems, respectively. Service discovery allows end systems to discover desired services on networks automatically, eliminating administrative configuration. We made four enhancements to the Service Location Protocol (SLP): mesh enhancement, remote service discovery, preference filters, and global attributes. These enhancements improve SLP efficiency and scalability, and enable SLP to better support new and advanced discovery scenarios. The SLP mesh enhancement (mSLP), remote service discovery, and preference filters are now experimental RFCs (Request for Comments). We expect that similar techniques can be applied to other service discovery systems. During the development of mSLP, we designed selective anti-entropy, a generic mechanism for high availability partial replication. Traditional anti-entropy only supports full replication. We enhanced it to support partial replication by allowing two replicas to selectively reconcile inconsistent data in a session. Web hotspots are short-term dramatic load spikes. We developed DotSlash, a self-configuring and scalable rescue system for handling web hotspots effectively. DotSlash works autonomously. It uses service discovery to allocate resources dynamically from a server pool distributed globally, and uses adaptive overload control to automate the whole rescue process. As a comprehensive solution, DotSlash enables a web site to build an adaptive distributed web server system on the fly, replicate application programs dynamically, and set up distributed query result caching on demand. DotSlash relieves a spectrum of bottlenecks ranging from access network bandwidth to web servers, application servers, and database servers. As part of DotSlash, we developed a prediction algorithm for estimating the upper bound of future web traffic volume, which is simple and effective for short-term bursty web traffic. This algorithm provides insight into characterizing traffic of web hotspots, and is useful for web server overload prevention.