Little’s Law: For Estimation Only

Dr. Rachel Traylor heard Datrium reference Little’s Law during their presentation at Storage Field Day last month. Little’s Law provides a way to smooth out some of the random variables in queuing theory into something more deterministic. As Dr. Traylor explains, this makes it idea for quick estimations, but it’s not a silver bullet for such a complex field of mathematics.

Dialogue: What do We Mean by Predictive Analytics?

At Storage Field Day this month, Dr. Rachel Traylor heard from StarWind. They developed a prediction algorithm to proactively detect drive failures. This led Dr. Rachel to call for a dialogue with her fellow mathematicians and storage experts, what do we mean when we say predictive analytics?

ML, AI, and Marketing: A Conversation with Dr. Rachel Traylor

Inspired by recent solutions seen at Tech Field Day presentations, Rich Stroffolino and Dr. Rachel Traylor discuss what actually is Machine Learning and Artificial Intelligence. They break down how having an algorithm doesn’t equal machine learning, and how to spot when marketing overreaches with the terms.

Commentary: White Papers Don’t Impress Me Much

In this post, Dr. Rachel Traylor looks at the current state of industry white papers. After surveying some recent papers promising architectural overviews and technical details, she mostly found them more akin to marketing materials due to the lack of citations and substance. She expresses this frustration with a little help from Shania Twain.

Commentary: Infrastructure Considerations for Machine Learning

In this post, Dr. Rachel Traylor looks at machine learning in the enterprise. The issue is machine learning takes vast swaths of varied data to produce useful results, something that might constrain a production database. Dr. Traylor looks at the Sky Infrastructure Actifio presented at Tech Field Day in September as a possible solution. This saves a “golden copy” of data, with changes saved incrementally, and able to serve up “virtual copies” much more efficiently for testing.

Commentary: High Level Data Filtration

Dr. Rachel Traylor looks at Ixia’s approach to real-time network visibility. This uses high level data filtration from a database of known bad actors to quickly eliminate large chunks of data from their analysis engine. This allows them to not have to process the entire firehose of network data and gives each successive analysis layer additional efficiency.