Sandeep Gopisetty, Eric Butler, et al.
IBM J. Res. Dev
Storage subsystems reside at the bottom-most layer of a contemporary data-center information technology (IT) stack. As with all other production software, the storage layer's embedded software (or firmware) must be constantly maintained in terms of upgrading it when new versions are released by hardware vendors. We study the problem of maintaining the currency of the storage software layer. We survey the typical processes that are used to keep storage firmware at recommended versions within a data center, and present an automated solution for this tedious and labor-intensive task. We present an approach to quantify the risk for continuing operations posed by stale firmware, a context-based approach for recommending target levels for devices with stale firmware, and an Upgrade Planner for rectifying the firmware of such devices. Our system has been deployed in the field using two approaches: as an operational tool used by hundreds of storage administrators within an outsourced IT context, and as an automated analytics component of an appliance-based approach within a maintenance services organization. We study the growth of the monitored infrastructure over two years. Finally, we show how our system drastically shrinks the time for an enterprise IT firmware upgrade cycle from days to minutes, and changes the nature of the complex task from a reactive to a proactive paradigm.
Sandeep Gopisetty, Eric Butler, et al.
IBM J. Res. Dev
Luis Bathen, German Flores, et al.
CVPRW 2019
Yuya Jeremy Ong, Jay Pankaj Gala, et al.
IEEE CISOSE 2024
Yu Deng, Kaoutar El Maghraoui, et al.
IBM J. Res. Dev