Keynotes

Event Processing in the Enterprise - a Waypoint on the Path to the Warehouse or the Launchpad for new Analytics Solutions (Industry Keynote)

Balan Sethu Raman - Microsoft

Download the keynote heretl_files/debs2012/img/icons/icon_slides.png.

The perception in the industry is that event processing solutions are either applicable to niche segments characterized by high velocity and volumes of data or as a mechanism to curate the data at the periphery to feed traditional data warehousing solutions.  There are a few enterprises in different industry segments that  have adopted a wider  role for event processing solutions to enable constant monitoring of the enterprise, managing the business and mining the assembled data. In this talk we will cover the lessons learned from those experiences and the challenges ahead for the research community.

Balan Sethu Raman was until recently a Distinguished Engineer with the Business Platforms Division in Microsoft. His career at microsoft spanned a number of product divisions. He was a member of the Windows team working on file systems. He led the development of the file system server product offerings during this period. In order to accommodate the needs of richer services like indexing, over vast amounts of data he pioneered an initiative to work with database team within Microsoft. This led to him transitioning into the Microsoft SQL server team. Working with Microsoft Office and Sharepoint services team he was responsible for introducing a variety of features into the database to facilitate document management oriented applications over relational databases. He subsequently worked with Microsoft research groups in conceiving and productizing a data stream processing product called Microsoft StreamInsight which is currently being used in a variety of event processing services and applications inside and outside Microsoft. He is at present working on plans to help develop solutions that can gather data from the physical world around us to yield insight into our activities.

tl_files/debs2012/img/persons/speakers/balan_sethu_raman.jpg

Actors Reloaded: From Scala Actors to Akka.

Martin Odersky - École polytechnique fédérale de Lausanne

Download the keynote heretl_files/debs2012/img/icons/icon_slides.png.

Scala had actors as a core library from very early on. Both event-based and blocking versions of actors were supported. For Scala 2.10, we redesigned the core concurrency abstractions for Scala and merged them with the distributed actor framework Akka. Starting with the JVM's fork-join framework, we build futures, promises and actors as higher-level constructs. One of the primary design objectives of these constructs was that they should be usable unchanged for both single-node concurrency and distributed systems. The talk reflects on our experiences with Scala actors and how they evolved to the new technology stack based on Scala and Akka.

tl_files/debs2012/img/persons/speakers/Martin_Odersky.jpg

Capturing Episodes: May the Frame Be with You

Dr. David Maier - Portland State University

Download the keynote heretl_files/debs2012/img/icons/icon_slides.png.

We are interested in detecting episodes in a data stream that are characterized by a period of time over which a condition holds, possibly with a minimum duration. For example, we might want to know whenever any router has a packet-drop rate above 0.1% continuously for more than a minute. Such episodes can be interesting in their own right for monitoring purposes, but they can also specify target regions for examination over the original or other stream. For instance, for each router-drop episode we detect, we might want to count the number of control messages the router received.

Current capabilities for data-stream management systems (DSMSs) include functionality, such as pattern-matching and windowed aggregates, that can help with detecting some kinds of episodes. We offer a third alternative, frames, which generalizes the other two. Frames are intervals that segment a data stream into regions of interest. In contrast to windows, frame boundaries can be data dependent, such as when the maximum and minimum values of an attribute diverge more than a certain amount or a running sum crosses a threshold. This talk introduces frames and their implementation in the NiagaraST DSMS. We then illustrate some of the advantages of frames versus windows, such as better characterization of episodes, adaptation to data and lower data rates to the client. The talk concludes with future directions for frame processing.

tl_files/debs2012/img/persons/speakers/David_Maier.jpg