Tamar Eilam  Tamar Eilam photo       

contact information

IBM Fellow
Thomas J. Watson Research Center, Yorktown Heights, NY USA
  +1dash914dash945dash2484

links


profile


Tamar Eilam joined IBM Research in 2000 and was recognised as an IBM Fellow in 2014. Tamar's research interest centers around the question of how to manage configuration complexity, and how to merge the gap between Development and Operations organizations (DevOps). Specifically, with the emergence of cloud environments, how do we deliver and continuously operate services with extreme agility, extreme performance, and reduced risk, and how do we gain insight to maximize the business outcome.

Tamar Eilam pioneering the concept of a "desired state based deployment and management". This principle centers around the idea of separating the "what" and the "how" in complex application deployments. A semantic description of a topology including software components, services, and infrastructure requirements is linked with automation to produce and converge to the desired state from the current state. This idea was the basis for a successful partnership with IBM's software group and the development of Rational's Deployment Planning and Automation offering (code name: Zephyr) and the virtual system editor technology in IBM Workload Deployer, SmartCloud Orchestrator, and Pure. Tamar led the the team that delivered many of the interesting features of these products: domain languages, validation and resolution to detect inconsistencies in desired state, and automatic workflow generation.

Tamar received her Ph.D. in the Computer Science Department in the Technion in Israel, where she studied algorithms in graphs and complexity, applied to communication networks. In addition, she worked as a summer intern in IBM Haifa Research Lab, on the cJVM (cluster JVM) project.

The Cloud, DevOps and Operational Analytics Department conducts research on how continuous deliver and operate cloud services with agility, optimization, and insight. Some of the research areas include:

Optimized placement of complex workloads composed of a network of resources based on a number of non-functional requirements.

Domain Specific Languages and tools for workload deployments, including testing for properties such as convergence, problem determination and debugging.

How to collect and analyze data in the cloud to gain insight.