[Links of the Day] 07/04/2020 : Incentivizing Innovation, Network Performance analysis, Neural Networks for embedded systems

• The Effects of Prize Structures on Innovative Performance : how to incentivize innovation? Well, the authors found that a winner-takes-all compensation scheme generates significantly more novel innovation relative to a compensation scheme that offers the same total compensation, but shared across the ten best innovations. However, like every psychological paper, you have to take it with a grain of salt.. reproducibility is always difficult.
• nfstream : Python package providing fast, flexible, and expressive data structures designed to make working with online or offline network data [github]
• Neural Networks on embedded systems : a good overview of the challenges and available neural network architectures for running on embedded systems.

[Links of the Day] 01/10/2019 : Capital venture relationships and investments, Neural Network normalisation Technic, Lossless data compression using deep learning

• The Dynamics of Venture Capital Relationships : Study of VC network of relationship. The authors found that having a deeper relationship leads to fewer, not more future co-investments. Moreover, deeper relationships lead to lower exit performance, even after controlling for endogeneity. Interestingly, deeper relationships first lead to lower performance and subsequently lead to a slowdown in the relationship intensity. Relationship effects are more negative for VC firms with less central network positions, and for deals made in “hot” investment markets.
• Cerebras : a new technique for normalizing hidden activations on neural networks. This allows researchers to greatly accelerated their training sequence without the need to be a Google or AWS with dedicated accelerators. [presentation] [git]
• Bit-swap : lossless data compression technique using deep learning.

[Links of the Day] 18/04/2019 : Decentralised networks made easy, Reload Linux system over ssh, Self Hosted push notification

• Octopus : Octopus allow you to streamline the creation of decentralized networks.
• reload.sh : Wipe, reinstall or restore your system from running GNU/Linux distribution. Via SSH, without rebooting. 
• Gotify : A self-hosted push notification service. But beware as worth noting that this will use a noticeable amount of power more than Google cloud messaging. Google/Apple and others have special deals with carriers that might make such solution cost prohibitive.

[Links of the Day] 08/11/2018 : large scale study of datacenter network reliability, What to measure in production, Failure Mode effect Analysis

• A Large Scale Study of Data Center Network Reliability : the authors study reliability within and between Facebook datacenters. One of the key findings is the growth in complexity, heterogeneity and interconnectedness of datacenter increase the rate of occurrence of unwanted behaviours. Moreover, this seems to be also a key potential limiting factor for world scale spanning infrastructure undergoing rapid organic growth.
• Understanding Production: What can you measure? : what do you need to monitor and measure in production. Very good summary of many blog post out there.
• Failure Mode Effects Analysis (FMEA) : once you start reaching a certain production scale and more stringent requirement kicks in ( unless you were unlucky enough to have them at the get-go). You might want to run a failure modes and effects analysis (FMEA) is a step-by-step approach for identifying all possible failures in a design, a manufacturing or assembly process, or a product or service. While it was mainly designed to address shortcomings in the manufacturing industry, it is still extremely useful for IT system analysis, especially when you want to prepare yourself pre-rollout of a chaos monkey like system.

[Links of the Day] 13/02/2018 : API aware network & security for containers with BPF, Machine Learning Data linter, Promise Theory & money

• Cilium : layer 3/4 networking management stack offering API-aware Networking and Security for Containers based on BPF. BPF is becoming the de facto tool for any high performance networking software out there. 
• data-linter : this is one of the must-have tool for ML. These show that the machine learning community is finally maturing out of the tinker age and into a more productivity age. This data linter offer a lightweight, automated sanity checking for ML datasets
• Promise theoretical analysis of money : an interesting look at money from the network technology point of view. It shows that the classic economic model might need to be revisited as monetary flow edge more toward the operational research/graph theory. 

[Links of the Day] 25/01/2018 : VPS comparison, Scale free network are rare and Data mining OCR PDFs

• VPS Comparison : this provides a very good overview of the different VPS provider out there. Obviously not complete, but hey, I would love to see a benchmark community driven website maintaining comparison for the different providers. Not just VPS, cloud / paas / lambda etc..
• Scale-free networks are rare : the ideal scale-free network does not really happen that often in the wild ( of the internet). Maybe its time to go beyond this concept and explore other more realistic avenues for real-world networks ( I'm looking at your distributed network of microservice) 
• Data Mining OCR PDFs : extracting info from PDF is a nightmare, it's even worse when you have to do OCR and I always considered that tabulation was a no go territory. But looks like somebody actually spent the effort to make it work and it's impressive.

[Links of the Day] 16/11/2017 : Sparse and dense array database, Rythm of memory, Routing over blockchain

• TileDB : manages massive dense and sparse multi-dimensional array data simply. This is a really good project as often there is no real support in existing database. 
• Rhythm of memory : the brain is a complex organ. And we just barely scratched the surface. Scientist discovered that part of the memory processing in the brain is segregated in different subcomponent that process information in parallel and at different speeds. This gives a glimpse of how the brain works and how it is able to store and access so much data at the various level of granularity fast.
• IPvPub : I really think there is something behind this concept. While I tend to be wary of the current trend of sprinkling blockchain everywhere. Using this technology for large-scale address resolution and routing can solve so many problems... Reduce reliance on DNS system in the age of lambda. What I really want to see is this integrated with a Lambda framework for simple exposure of service endpoints.

[Links of the Day] 06/10/2017 : HPC routing topology on dependency graph, Arm network stack development, Multicore graph processing

• Routing on the Channel Dependency Graph : the author aim are providing a toolset for topology calculation for HPC network. 
• Arm for network stack developers : arm is trying to slowly move up the stack into the data centre world. For that, it needs to address one of its main limitation: IO. This slide deck describe the current effort to tackle network stack limitation ( and support RDMA ) as well as providing pointer where ARM or devs can push the Linux stack further. As Linux is pretty much the only viable software stack for such hardware infrastructure in the datacenter. 
• Multicore graph processing : this present a very good overview of the Multicore graph processing problem, solution landscape and where it's heading. Graph problems are essential to solve in the domain of social network modelling as well as for items recommendation or search and website ranking. [slides]

[Links of the Day] 04/10/2017 : Tensor Flow Extended , 5d Torus , Machine learning influencing hardware designs

• TFX : Google paper describing it's tensor flow based platform for production scale deployment. It is impressive how they created a platform that delivers robust TensorFlow-based learners and support for continuous training and serving with production-level reliability. [video]
• 5d torus : when your 3d torus fabric is not enough :) . The authors demonstrate how using 6 port NIC they are able to achieve such feat. Its domain of application is still be limited to HPC environment as only tailored applications can leverage such network topology. However, there might be potential in the machine learning domain. There is only a future for such approach if they are able to demonstrate significant acceleration for popular problems ( ML ). And make is fungible for cloud-like deployment. I.e. you need to be able to easily partition a set of nodes connected by such network and share it in multitenancy model. [slides] [thesis]
• Machine Learning and the Implications for Computer System Design : Jeff Dean talk at hot chips 2017 conference. Jeff really really show the current influence of ML on the present and future hardware architecture. 

[Links of the Day] 27/07/2017 : Russian Information Warfare, Public Cloud Economics, Impact of slowness in ultra fast networks

• Russian Information Warfare Handbook : In depth description of Russian information war machine and how it affects the rest of the world. Some suggestion on how to fight it, however, I feel that the internet and the current social networks model heavily bias in favour of propaganda exploitation. Moreover, with the advance of machine learning, it becomes easy to identify key nodes in the network to attack in order to achieve maximum disinformation reach. 
• Usage Patterns and the Economics of the Public Cloud : paper looking at the current cloud economic and cloud customers. It seems that most customers have a steady or mild variation usage with the obvious outlier ( probably academia or other heavy HPC / Machine learning job) 
• To slow or not? : the authors look at the impact of slowing down (hardware error, software bug, or simply laws of physics)  next gen network on data centres and application when we start to reach 1 digit microsecond latency. There is a wide range of implication, from technical to legal as well as ethical. Sadly most of them tend to stay unanswered until a problem or conflict arise. 

[Links of the Day] 11/05/2017 : Google - Push on green , Tensor flow in Datacenter , TCP congestion protocol

• In-Datacenter Performance Analysis of a Tensor Processing Unit : How custom deep learning hardware behave in real datacenter, implication and gain associated with the use of such custom solution.
• Push on Green : Great article on google roll out policy and process. A lot of common sense, and also some less common but as important. This is a great read for anybody involved in software delivery and especially if you are aiming at having an efficient CI/DI system. 
• BBR : google congestion protocol for maximising bandwidth usage. It's new TCP scheduling algorithm to fight buffer-bloat at the TCP level. Since the majority of internet traffic is TCP, wide adoption would cause a big improvement. TCP scheduling only affects outgoing packets

[Links of the Day] 05/04/2017 : Large network resilience, Distributed Systems, Machine Learning & Bayesian reasoning

• Understanding resilience in large networks : Thesis looking into the analysis of robustness in large interconnected networks. It looks at how disturbance propagates through nodes in networks and the characteristics that help these network to handle such shock and recuperate from it. 
• Notes on Theory of Distributed Systems :  Yale distributed system course note. 
• Bayesian Reasoning and Machine Learning : Book that does an excellent job at teaching machine learning that doesn't assume an extended background in undergraduate calculus and linear algebra of the reader.

[Links of the Day] 23/12/2016 : Microsoft Configurable cloud (with fpga), Open Pilot OSS driving software, Deep learning is all about rigor

• Microsoft's Production Configurable Cloud : built in custom nic + fpga for highly configurable and dynamic network stack in Microsoft DC. The work is really impressive. It demonstrate how pervasive FPGA and customization hardware will be in future datacenter. 
• Open Pilot : open source driving agent providing Adaptive Cruise Control (ACC) and Lane Keeping Assist System (LKAS) for Hondas and Acuras. This is a really interesting solution and I wonder how fast other company will start to leverage or opensource their own solution in order to accelerate adoption. However without extremely strong verification and proof system ( formal method ) it will be extremely hard ( and illegal probably ) to deploy such software at this stage. 
• Nuts and Bolts of Building Deep Learning : Andrew Ng reiterated at NIPS2016 that there is no secret AI equation that will let you escape your machine learning woes. All you need is some rigor. [video]

[Links of the Day] 28/11/2016 : Earth Computing network fabric, CS video courses, Go app tracing

• Earth Computing Network Fabric : event based protocol for datacenter that target specifically datacenter as it eliminate the need for heartbeats and timeouts. The protocol relies on recoverable atomic token to deliver deterministic in order communication. To some extend they are proposing to move back to latices system where each server are a node within the network and act as router for messages. This eliminate switch requirement and looks really neat. However adoption might be difficult due to the ubiquitous Ethernet hardware and also the need to change the underlying communication protocol. Last but not least I do not really know how to efficiently secure and trust messages on such network. [slides] 
• Computer Science video courses : Extensive collection of links to CS courses ranging from introductory to expert in pretty much the full scope of CS subject (DB, distributed systems, etc..) 
• Appdash : Application tracing system for Go, based on Google's Dapper.

[Links of the day] 23/05/2016: computational law , EU data protection regulation & 2M packets/s on AWS

• CodeX : Stanford R&D lab of computational law — the branch of legal informatics concerned with the automation and mechanization of legal analysis. I really think that we are getting to the point where every single IP and Assets will be handled/ represented by an AI and where the actual ownership of assets will be dynamically changing at the speed of computation Accelerando style. [Youtube Videos]
• 2 Millions packets seconds on AWS : leveraging SRIOV tech on AWS to process packet at high speed.
• EU General Data Protection Regulation : a great document dissecting the upcoming European General Data protection Regulation directive. This directive cover a very wide range of legislation, from right to be forgotten, data protection to compliance.  

[Links of the day] 25/04/2016: Approximate computing benchmark suit, Awesome Network analysis and complex network book

• AxBench : This benchmark suite offer 29 carefully selected benchmarks to evaluate the approximate computing capability of a system stack. [paper]
• Awesome Network Analysis : list of resources to construct, analyze and visualize network data.
• Large Scale Structure and Dynamics of Complex Networks : fantastic book providing a unifying picture of complex networks.

Network of U.S. political blogs by Adamic and Glance (2004) (preprint)

[Links of the day] 16/03/2016: userspace TCP stack, pebcak, neural net

• mTCP : Highly Scalable User-level TCP Stack for Multicore Systems, really cool project released back in 2014. The really interesting bit is that they now support DPDK natively which allow for massive performance improvement while bypassing the kernel stack altogether. [github]
• The problem is between the chair and the keyboard : how do you patch users ? A Taxonomy of Attacks and a Survey of Defence Mechanisms for Semantic Social engineering Attacks
• Neural Networks Demystified : series of video explaining neural networks

[Links of the day] 26/02/2016 : Usenix Fast 16 , FPGA liberouter and Event delivery at spotify

• Fast 2016: all of Usenix Fast 2016 goodness available in one place.Interesting to see that we start to see the emergence of storage systems optimized for time series [BTrDB] . Also note the always interesting report on failure rate (this time for flash)
• Liberouter : really cool project using FPGA to deliver hardware acceleration of network security and monitoring tools. 
• Event Delivery at spotify : part 1 of a series of blog post on event monitoring and management system used at spotify.

[Links of the day] 09/02/2016: Run openstack in containers, Flocon network security conf, Agent motion book

• Kolla : run Openstack in containers for ease of management and operational flexibility.
• Flocon 2016 presentation : Network Situational Awareness for large-scale network flow analytics.
• Space & motion of communicating agents : CS thesis looking at leveraging bigraph to model the space , motion of communicating agents. This essential as for the pas decades system mainly focused on getting the itnernal communication of entity right. But with the upcoming Punctuated equilibrium of IoT we external communication systems in an non static world will become critical.

Links of the day 07/12/2015: Netflix Streaming architecture, Silicon Valley Satire, FLOSS doom checklist

• 1/3 of internet Traffic : How Netflix Direct architecture was designed to route a third of the world internet traffic. 
• How to tell if a FLOSS project is doomed to FAIL : quick check list to get an indication if an opensource project is doomed on not.
• Iterating Grace : satire about a Silicon Valley programmer named Koons Crooks, who has a spiritual awakening that leads him on a pilgrimage to a volcano in Bolivia. Living in a yurt, he spends his days pondering the transcendent wisdom of tweets by top venture capitalists before apparently being trampled to death by vicuñas. Just read it.