In this iteration of Gestalt Server News:
– A look at AMD and Intel’s surprisingly competitive server platforms
– HyperGrid’s on-demand on-site solution
– How VMs are still more portable than Docker containers
Changes in AI: Solving Checkers
Checkers is the game I played to kill time waiting for tables at restaurants. But solving checkers turns out to be a fascinating exercise. Recently, Alphabet’s AlphaGo team has made a lot of headlines with their neural network-based ability to beat human Go masters. But Ray Lucchesi looks back at earlier days trying to solve checkers with much more limited hardware and fundamentally different approaches.
Is DevOps a Disaster? The On-Premise IT Roundtable 2
Podcast: Play in new window | Download (Duration: 22:35 — 20.7MB)
Subscribe: Apple Podcasts | Android | Google Play | Stitcher | RSS | More
Moderator Stephen Foskett poses a completely non-controversial question: is DevOps a load of crap? Does DevOps just turn into NoOps? What are these darn kids doing with our infrastructure? The roundtable debates all these questions and more.
Hardware vs. Software Innovation
Ray Lucchesi recently weighted in on where innovation is occurring in IT, hardware or software. He considered Dell EMC’s decision to kill their DSSD NVMe storage device, and frames it in the continuing debate. Ray thinks it’s further evidence that we are in a software innovation cycle. As further evidence, recent releases by both Excelero and E8 Storage. Both are using commodity hardware to achieve high level performance, over 4 million IO/sec with ~120 to ~230µsec response times. It’s an interesting discussion, and Ray gives both sides their due.
TPU and hardware vs. software innovation (round 3)
Ray Lucchesi of RayOnStorage Blog comments: At Google IO conference this week, they revealed (see Google supercharges machine learning tasks …) that they had been designing and operating their own processor chips in order to optimize machine learning. They called the new chip, a Tensor Processing Unit (TPU). According to Google, the TPU provides an order of magnitude more […]
Faster Docker initialization through Slacker snapshots & NFS storage
Ray Lucchesi of RayOnStorage Blog comments: Just got back from EMCWorld2016 this week but on the way there and back I was perusing the FAST’167 papers. One of the papers I read (see Slacker: Fast Distribution with Lazy Docker Containers, p. 181) discussed performance problems with initializing Docker container micro-services and how they could be […]
Surprises from 4 years of SSD experience at Google
Ray Lucchesi of RayOnStorage Blog comments: Flash field experience at Google In a FAST’16 article I recently read (Flash reliability in production: the expected and unexpected, see p. 67), researchers at Google reported on field experience with flash drives in their data centers, totaling many millions of drive days covering MLC, eMLC and SLC drives with a […]
Rubrik has a better idea for VMware backup
Ray Lucchesi of RayOnStorage Blog comments: Rubrik is out to change that and has taken a different tack to doing backup for VMware. Most backup packages are horendous to configure, requiring a multitude of jobs to be scheduled, backup proxies to be configured, backup servers, catalog service/search services, etc. All that just to backup a […]
Coho Data, the packet processing squeeze and working set exploits
Ray Lucchesi of RayOnStorage writes: Andy’s always a joy to talk with and this time was no exception. Andy started out talking about the speed of networking and what it meant for network packet processing time. He showed a chart with network speeds on the horizontal axis and packet processing time (in nsec) on the […]
Moore’s law is still working with new 2D-electronics, just 1nm thin
Ray Lucchesi of RayOnStorage Blog comments: This week scientists at Oak Ridge National Laboratory have created two dimensional nano-electronic circuits just 1nm tall (see Nature Communications article). Apparently they were able to create two crystals ontop of one another, then infuse the top layer with sulfur. With that as a base they used standard scalable […]