- Aphyr Scala Day 17 : Aphyr breaks database for a living and then talks about it :)
- Sia : a Blockchain-based marketplace for file storage, the really attractive thing is the cost comparison of SIA vs public cloud system. Which is between a tenth to a hundredth time cheaper than S3 or other similar solution. I would be curious to see the performance thought.
- How brains are built: High-level overview of principles of computational neuroscience.
Wednesday, April 26, 2017
Monday, April 24, 2017
[Links of the Day] 24/04/2017 : Data Center Perf Index, Physical Limits of Computing And DNS as Code
- Data Center Performance Index : Performance index aiming at providing a reliable idea of the performance and efficiency of the datacenter. It primarily focuses on Availability, Efficiency and Environmental impact. This effort is lead by Dean Nelson, Uber Had of computing.
- Physical Limits of Computing : A look at the limitation of compute from a physicist point of view. It seems that some limitations are fundamental and require a new and different approach in order to create compute device that goes around these limitations.
- DnsControl : system built to manage DNS systems. To some extent, it looks like a terraform for DNS where you can plug multiple DNS backend providers. This allows you to deploy and distribute your DNS infra in an agnostic way across multiple cloud provider. [github]
Friday, April 21, 2017
[Links of the Day] 21/04/2017 : HPC 2017 trends, Docker cheat-sheet, Incident response best practice
- Current Trends in High-Performance Computing and Challenges : Jack Dongarra annual HPC review, It's amazing how the chinese progressed. They literally took over the top 500 in less than ten years. And now they dominate using homegrown chips and network fabric. [slides]
- Docker cheatsheet : 'nough said.
- Increment - On-call : New magazine providing article on how to scale companies. Each edition focus on a different topic. For the inaugural issue, they focus on industry best practices around on-call and incident response.
Wednesday, April 19, 2017
[Links of the Day] 19/04/2017 : AMD ROCm GPU open platform, Weak Memory Models concurrency report, SSH server for distributed infrastruscture
- ROCm : this slide deck give an overview of the AMD ROCm open platform for GPU computing exploration. They are really pushing to become the open source standard for the GPU industry battling against NVIDIA supremacy in the domain. It looks like they are making really good progress and I would be curious to see how this progress when combining with their Ryzen CPU.
- Concurrency with Weak Memory Models : this is a really good report on the state of memory models in hardware and software. It provides a wide spectrum overview of Hardware and Software concurrency model and approaches as well as the future direction in the domain.
- Teleport 2 : a modern SSH server designed for teams managing distributed infrastructure. [github]
Monday, April 17, 2017
[Links of the Day] 17/04/2017 : Pedis Redis Clone, Serverless framework, Deep learning best practices
- Pedis : Redis Compatible NoSQL datastore using the Seastar Framework. It's interesting to see that on the single thread benchmark Redis and Pedis are on par while it Redis gets smoked on 8 thread benchmark. However on a side note, the author should probably have chosen another name for project.
- serverless : Serverless Framework with serverless architectures using AWS Lambda, Azure Functions, Google CloudFunctions [github]
- Best Practices for Applying Deep Learning to Novel Applications : this is pretty much a must read for machine learning expert using deep learning. This report decomposes deep learning project in phases and provides best practice for each phase.
Friday, April 14, 2017
- OpenFabrics Workshop 2017 : Some interesting talk this year at the open fabric conference:
- uRDMA : Userspace RDMA using DPDK. This opens up a certain amount of possibility, especially for object storage solution. [Video , Slides, github]
- Crail : Using urdma above to deliver accelerated storage solution for Apache big data projects [Slides, github]
- Remote Persistent Memory: I think this is the next killer app for RDMA. If Intel doesn't jump onto it and deliver a dpdk like solution. [Video, Slides]
- On Demand paging: slowly the tech is crawling its way up to upstream acceptance. While on-demand paging introduces a certain performance cost. It also allows a greater flexibility in consuming RDMA. One of the interesting aspects that nobody mentioned yet is how this feature could be used with persistent memory. I think that there is some good potential for p2p NVM storage solution.[Video, Slides]
- Containerd : Containerd move to github, the docker "industry standard" container runtime is also reaching its v.0.2.x release. [github]
- Category Theory : If you are into functional programming and Haskell. This is a must read book for you.
Wednesday, April 12, 2017
[Links of the Day] 12/04/2017: Linux Perf tools, libp2p and Contagion of Information in Social Media
- Perf Tools : miscellaneous collection of in-development and unsupported performance analysis tools for Linux ftrace and perf_events.
- Contagion of Information in Social Media : The authors look at how information spread on social media ( twitter ). The authors model contagion behaviour in the hope to create effective defences against "fake news" and other propaganda. However, to some extend the research can also be used to optimise the spread of such malicious information.
- libp2p : really cool network stack ( used by IPFS) that tackle a lot of the nitty gritty detail of p2p applications. It should allow devs to focus on the actual value of their p2p apps rather than the technical underlying problems of p2p itself. [github]
Monday, April 10, 2017
[Links of the Day] 10/04/2017 : Loopy , Distributed execution engine and Yet another distributed ledger algo
- Loopy : Fantastic tools for explaining and describing complex system interaction. It's easy to use and even easier to get the message across [github]
- Ray : experimental distributed execution engine replicating code across multiple workers. Written in python it leverages object store and distributed task execution to achieve parallelism. I really wonder if it wouldn't have been better to code Ray using AWS lambda and S3.
- Algorand : yet another distributed ledger. The approach proposes to eliminate the segregation of actors in the ledger system. No more miners and users, everybody is an equal participant. Moreover, it relies on a new for of Byzantine agreement ( need a TLA+ proof to really feel comfortable with that) and cryptographic selection algorithm for selecting leader (verifier) of the ledger process execution. This is a rather interesting paper and I will try to produce a short summary of it if I find the time.
Friday, April 07, 2017
[Links of the Day] 07/04/2017 : TensorFlow Example, Systems in the microseconds era and blockchain distributed direct democracy
- Naked Tensor : Google tensor flow bare-bone example.
- Attack of the Killer Microseconds : Hardware ( especially storage ) is entering the micro or sub-micro second era. This has far and wide ranging implication. And system designer needs to rethink the existing stack that was designed for the millisecond era. It looks like we are entering an era where the software stack is not the bottlenneck.
- Cicada : Distributed secure proof of work blockchain combined with a privacy guaranteed ID system. The creator of cicada aim at enabling distributed direct democracy and decentralised application platform. This is worthy goals, however, the creator forgot that direct democracy tend to fail as the majority of the population is not interested or knowledgeable enough in the problem they will be asked to vote on.
Wednesday, April 05, 2017
[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.
Monday, April 03, 2017
[Links of the Day] 03/04/2017 : Conway's Game of life Clock, Human-Bot social interaction, SQL time series DB
- Digital clock in Conway's Game of Life : I can't even start to comprehend how you can design this. But this is beyond cool.
- Online Human-Bot Interactions: Detection, Estimation, and Characterization : An analysis of bots on socials network (twitter). I think we need a reverse Turing test. When a robot can detect when they talk to a human.... Reverse captcha to weed out that pesky meat-bag from meddling from our robotic overlord affairs.
- Timescale : SQL compatible time series database. Another competitor for Influxdb. Let's just say that the clustering feature will make or break it as Influxdb has some serious issue there [github]