Happy first birthday to the VMware Technical Journal! We are very pleased to have seen such a positive reception to the first couple of issues of the journal, and hope you will find this one equally interesting and informative. We will publish the journal twice per year going forward, with a Spring edition that highlights ongoing R&D initiatives at VMware and the Fall edition providing a showcase for our interns and collaborators.
VMware’s market leadership in infrastructure for the software defined data center (SDDC) is built upon the strength of our core virtualization technology combined with innovation in automation and management. At the heart of the vSphere product is our hypervisor, and two papers highlight ongoing enhancements in memory management and I/O multi-pathing, the latter being based upon work done by Fei Meng, one of our fantastic PhD student interns.
A fundamental factor in the success of vSphere is the high performance of the Tier 1 workloads most important to our customers. Hence we undertake in-depth performance analysis and comparison to native deployments, some key results of which are presented here. We also develop the necessary features to automatically manage those applications, such as the adaptive task management scheme described in another paper.
However, the SDDC is much more than just a large number of servers running virtualized workloads—it requires sophisticated analytics and automation tools if it is to be managed efficiently at scale. vCenter Operations, VMware’s automated operations management suite, has proven to be extremely popular with customers, using correlation between anomalous events to identify performance issues and root causes of failure. Recent developments in the use of graph algorithms to identify relationships between entities have received a great deal of attention for their application to social networks, but we believe they can also provide insight into the fundamental structure of the data center.
The final paper in the journal addresses another key topic in the data center, the management of energy consumption. An ongoing collaboration with Boston University, led by Professor Ayse Coskun, has demonstrated the importance of automatic application characterization and its use in guiding scheduling decisions to increase performance and reduce energy consumption.
The journal is brought to you by the VMware Academic Program team. We lead VMware’s efforts to create collaborative research programs, and support VMware R&D in connecting with the research community. We are always interested to hear your feedback on our programs, please contact us electronically or look out for us at various research conferences throughout the year.
VMware Academic Program