USENIX Lecture: Next Generation Storage Architecture

Image depicting Cambridge Computer services

Next Generation Storage Architectures dominated the conversations of a five-hour gathering of IT professionals from across the UNT System for a USENIX lecture. People interested in storage architectures had a full agenda as they met April 10, 2017 in the University Union, Room 314A. Since 1975, the USENIX Association has brought together the community of engineers, system administrators, scientists, and technicians working on the cutting edge of the computing world. The USENIX conferences have become the essential meeting grounds for the presentation and discussion of the most advanced information on the developments of all aspects of computing systems. 

Dr. Philip Baczewski introduces the USENIX speaker Jacob FarmerPhilip Baczewski, senior director, University Information Technology, provided opening remarks noting that at the start of his career, the baseline standard for data storage was the 80-byte IBM punch card. Over the years capacities have grown through kilobytes, megabytes, gigabytes, and terabytes, with management of petabytes of data being the norm in today's data centers. Large commercial online providers, such as Google and Amazon, now manage exabytes of storage and the size of all the world's digital information is projected to grow from 4.4 zettabytes in 2015 to 44 zettabytes by 2020.

He noted that the job of many IT professionals these days involves managing "big" data or supporting the use of large collections of information within the IT infrastructure. Baczewski then introduced the featured speaker Jacob Farmer from Cambridge Computer Services.

Starfish company, Jacob Farmer, founderAbout Jacob Farmer: For the last two decades Jacob Farmer has been a lecturer and thought leader on data storage technologies for the Usenix community. He has taught tutorials, led guru sessions, and chaired panels for a number of Usenix conferences and events. In his day job, Jacob is the chief technology officer for Cambridge Computer, a position he has held for 25 years. Cambridge has a unique business model, like that of a broker or agent, helping companies define and refine their data storage, cloud and virtualization strategies. He also is the founder and "chief evangelist" for Starfish Storage, a software company automating the life cycle of institutional file collections.

The program as announced included the following two main presentation topics.

UNT staff and faculty attend USENIX presentationStorage Marketecture: What is Really New in Data Storage and Where is it All Going?
There is always something new in data storage ... or is there? Is software defined storage new? Is hyper­convergence anything more than virtualizing the array controller? Object stores aren't new. They have been around forever. Even erasure codes date back to the 1960s. This session is a crash course in the new stuff that is actually changing the data storage landscape. We trace the evolution of modern storage systems from lowly hard drives and file systems to modern cloud-scale distributed storage architectures. We answer the questions what is really new, how does the evolution of hardware influence best practices in storage, and how is any of this going to get the job done faster, easier, or more economically?

A Crash Course on Object Storage
Next Generation Storage Architectures for High Performance, Archives, and Everything In Between The storage industry is hyping object storage again! The last time was about 15 years ago when object­based file systems first came to market as solutions for low latency, high performance computing. This time, the hype is around RESTful object stores like S3 and SWIFT which are used predominantly for high latency, low throughput applications. What could these use cases possibly have in common as to require storage objects?

The session hit the reset button on object storage. Starting with definitions of object and object store and describing key concepts in object storage: deduplication, redundancy models, fine-grained v. coarse grained, nesting of objects, shipping manifests, and content addressing, it was attended by UNT faculty, staff and students. Farmer also looked at various implementations of object storage in enterprise storage devices. The lecture concluded at 5 p.m. with a segment that illustrated various ways in which object storage enables tiered storage, archiving and data life cycle management; the attendees stayed after the lecture for an additional hour of informal networking.

Editor's Note: Please note that information in each edition of Benchmarks Online is likely to change or degrade over time, especially links to various websites. For current information on a specific topic, search the UNT website, UNT's UIT Help Desk or the World Wide Web. Email your questions and comments to the UNT University Information Technology Department or call 940-565-4068.