My time as an intern (thus far)

My time as an intern (thus far)

It has been over three months since I started as a marketing intern at Cloudflare. Even before joining the Cloudflare team, I enjoyed reading the technical blog posts about Cloudflare’s use cases and solutions, as well as the inclusive and creative culture. Educating the world about the threats we face on the Internet is something that I found truly valuable. I figured that I would give my own spin on what it’s like to join and work at Cloudflare by writing a blog post too.

Chapter 1: The Path towards the Orange Cloud

My time as an intern (thus far)
Photo by alexander milo / Unsplash

Before starting as a freshman in university, I created an online portfolio for my photography. It has been a passion of mine for about 9 years. I tried a multitude of platforms but none afforded me the aesthetic control that I wanted. The only solution was to build and host my own site. I started learning HTML/CSS, a bit of JavaScript and jQuery and so on.

This led to me using the Koken CMS, and hosting it on DigitalOcean in a Docker container. What was left was SSL/TLS encryption and a CDN (my friends in Europe find the loading times unbearable). Continue reading

Continuum: a platform for cost-aware low-latency continual learning

Continuum: a platform for cost-aware low-latency continual learning Tian et al., SoCC’18

Let’s start with some broad approximations. Batching leads to higher throughput at the cost of higher latency. Processing items one at a time leads to lower latency and often reduced throughput. We can recover throughput to a degree by throwing horizontally scalable resources at the problem, but it’s hard to recover latency. In many business scenarios latency matters, so we’ve been seeing a movement overtime from batching through micro-batching to online streaming.

Continuum looks at the same issues from the perspective of machine learning models. Offline (batch) trained models can suffer from concept drift (loss of accuracy over time) as a result of not incorporating the latest data. I.e., there’s a business cost incurred for higher latency of update incorporation. Online models support incremental updates. Continuum determines the optimum time to retrain models in the presence of incoming data, based on user policy (best effort, cost-aware, or user-defined). There’s some great data here about the need for and benefit of continual learning, and a surprising twist in the tale where it turns out that even if you can afford it, updating the model on Continue reading

Cumulus Networks Open-Ended NCLU Net Example Command

NCLU is the always helpful Network Command Line Utility. It’s a command interface for our products and platforms that’s designed to provide direct, simple access to network configuration information. Thus, NCLU supports both inspection and modification of Cumulus Networks configuration data. Better yet, NCLU is easy to customize for local environments and naming conventions using its net example facility.

NCLU Overview

In general, NCLU enables users at the command line to learn about current configurations, and make changes or additions to such configurations. NCLU reports on Interfaces and can provide information about IP addresses, VLANs, Access controls, Trunking, STP, and more. At the routing level, NCLU provides information about Border Gateway Protocol (BGP) and Open Shortest Path First (OSPF) routing protocol settings and configurations. NCLU also offers information about services, including hostnames, NTP (Network Time Protocol), Timezone, and so on.

NCLU also includes comprehensive, context-sensitive help. Starting with the basic net command, users can learn about the various sub-commands available to them. Similarly, entering net <sub-command-name> provides help for that specific sub-command. This is how Cumulus (and other forms of) Linux delivers help information for users of complex commands like net.

In addition, NCLU commands provide control over configuration staging, Continue reading

BrandPost: SD-WAN: Optimizing User Experience for Managed Cloud-hosted Applications

Digital Transformation touches the network Applications are moving to the cloud – a lot of them. IDC estimates that by 2020, 90% of enterprises will use multi-cloud, i.e. multiple pub­lic and private cloud services and platforms, to support their ever-expanding application requirements. Service providers have a unique opportunity to leverage a high-performance managed SD-WAN solution to deliver the best possible cloud connectivity and therefore the best “quality of experience” to enterprise users.A new breed of applications with different needs Applications require a different quality of experience based on business priority, geography and security considerations, and consequently, they must be handled accordingly across the WAN. Some trusted business applications like Office 365, Skype and SAP can be sent directly to the internet with confidence, while recreational applications such as Facebook and Twitter might require higher security controls in order to meet enterprise security and/or compliance requirements. Simply put, the SD-WAN internet connectivity option must include the ability to differentiate traffic based on each application to ultimately enforce granular security policies.To read this article in full, please click here

Cumulus Linux Automation with Standard Linux Tooling

One thing’s for sure: The world of networking and networking administration is quickly changing. Part of this change is an evolution from old-school, proprietary centralized networking to more open options. This evolution has several different effects on the way network designers, administrators and engineers design and operate the network. This blog will focus on the different options available for modern automation, and how the Cumulus Linux approach provides the greatest amount of flexibility.

Breaking the Stranglehold

It wasn’t too long ago that the few big networking vendors had an almost unbreakable grip on organizational networking implementations, and correspondingly, with the way these implementations were managed. For most, this included the configuration of the various types of networking equipment using a command-line interface (CLI) and proprietary commands. Automating these types of solutions most often required either an offering developed by the vendors themselves, or the use of an application programming interface (API) written to interface with their products.

The question is whether this was a good thing or not. Generally, vendor-specific solutions have their advantages because they’re able to interface closely with the specific device code and take advantage of communications between the device coding team and the tools coding team.

Continue reading

Chip-cooling breakthrough will reduce data-center power costs

Traditional passive heatsinks affixed to microprocessors for cooling don’t work well enough for today’s high-speed computations and data throughputs and should be junked, says a group of mechanical engineering researchers.A better option, they say, are "spirals or mazes that coolant can travel through" within tiny channels on the actual processor. That technique could massively improve efficiency, says Scott Schiffres, an assistant professor at Binghamton University in New York, in an article on the school's website. The school has developed this new method for cooling chips.To read this article in full, please click here

Chip-cooling breakthrough will reduce data-center power costs

Traditional passive heatsinks affixed to microprocessors for cooling don’t work well enough for today’s high-speed computations and data throughputs and should be junked, says a group of mechanical engineering researchers.A better option, they say, are "spirals or mazes that coolant can travel through" within tiny channels on the actual processor. That technique could massively improve efficiency, says Scott Schiffres, an assistant professor at Binghamton University in New York, in an article on the school's website. The school has developed this new method for cooling chips.To read this article in full, please click here

IDG Contributor Network: Zero-trust security adds necessary ingredients

Today's threat landscape consists of skilled, organized and well-funded bad actors. They have many goals including exfiltrating sensitive data for political or economic motives. To combat these multiple threats, the cybersecurity market is required to expand at an even greater rate.The IT leaders must evolve their security framework if they want to stay ahead of the cyber threats. The evolution in security we are witnessing has a tilt towards the Zero-Trust model and the software-defined perimeter (SDP), also called a “Black Cloud”. The principle of its design is based on the need-to-know model.The Zero-Trust model says that anyone attempting to access a resource must be authenticated and be authorized first. Users cannot connect to anything since unauthorized resources are invisible, left in the dark. For additional protection, the Zero-Trust model can be combined with machine learning (ML) to discover the risky user behavior. Besides, it can be applied for conditional access.To read this article in full, please click here

IDG Contributor Network: Zero-trust security adds necessary ingredients

Today's threat landscape consists of skilled, organized and well-funded bad actors. They have many goals including exfiltrating sensitive data for political or economic motives. To combat these multiple threats, the cybersecurity market is required to expand at an even greater rate.The IT leaders must evolve their security framework if they want to stay ahead of the cyber threats. The evolution in security we are witnessing has a tilt towards the Zero-Trust model and the software-defined perimeter (SDP), also called a “Black Cloud”. The principle of its design is based on the need-to-know model.The Zero-Trust model says that anyone attempting to access a resource must be authenticated and be authorized first. Users cannot connect to anything since unauthorized resources are invisible, left in the dark. For additional protection, the Zero-Trust model can be combined with machine learning (ML) to discover the risky user behavior. Besides, it can be applied for conditional access.To read this article in full, please click here

1 … 1,449 1,450 1,451 1,452 1,453 … 3,897