Jay Fenton


Founder & CTO at Savvi

Jay Fenton grew up on a steady diet of grep, awk and sed, discovering UNIX at a very early age. This, coupled with a near-obsession with the phone system early in life saw Jay’s career inevitably span the gamut of telecoms, networks, and polyglot software development at every layer of the stack.

Cutting his teeth as a sysadmin and running and developing OSS/BSS systems in the mid-90’s for major carriers, he went on to develop software for next generation network management systems vendor, Riversoft, before starting his entrepreneurial career in 2003. After a brief stint in the HR space before deciding it was hellishly boring, he founded several small startups in the telecoms space, working on everything from low-level transcoding solutions on exotic processors, to innovative conference calling platforms written in Erlang, and everything in between.

His latest endeavour – Savvi Inc. – is stealthily revolutionising how complex networks are managed at truly massive scale, corralling the messy data endemic to telecoms, and intimately connecting the network engineer directly to the network at large.

It is said that at night his dreams are encapsulated in SNMP packets, which is not necessarily a good thing.

YOW! 2015 Melbourne

The Pathologies of Big “Messy” Data in Telco


In this talk, Jay will share his experiences of dealing with the realities of telecoms data at scale. The tools he uses are broadly applicable outside of the domain.

Modern networks are today still being managed with tools that were built over a decade ago. It sounds crazy, but it’s true.

Nowhere is this more ludicrous than in the world of telecoms operators, where Big Data was all but invented. The technologies coming out of the social- and web-scale space are all well-and-good when the data is well formed, and understood, and where the organisation is relatively flat. But in telco, it’s a different world.

ITIL, eTOM and the Common Information Model whilst widely adopted, have done nothing to improve the realities of dealing with the typical data found in the telecoms domain. It’s messy, voluminous, fragmented, duplicated, and at times seems almost purposefully obfuscated! As our networks get larger, as access technologies proliferate, as the products get more complex and the accumulated cruft of legacy products takes a long time to shake off, the way we manage network telemetry pipelines and the data that makes sense of it simply must evolve.

In this talk, Jay will share his experiences spanning Splunk® and it’s Unix Pipe Philosophy “SPL” language, coupled with Titan, and several other tools and technologies to make sense of that data and unravel the mess, at scale.