To improve network throughput, scrap the ones-and-zeros system

Abandoning a signal’s traditional binary ones and zeroes will improve throughput in fiber, a Dutch PhD researcher says. By increasing the number of bit states from the traditional two — the one and zero — to four or eight, adding more light colors, and changing clock rates for each wavelength, throughput will be significantly increased, he claims.“A higher number of levels implies more information per symbol, so four levels contain two bits of information, and eight levels contain three bits,” says Robbert van der Linden in an article on the Eindhoven University of Technology Department of Electrical Engineering website.To read this article in full, please click here

Technology Short Take 98

Welcome to Technology Short Take #98! Now that I’m starting to get settled into my new role at Heptio, I’ve managed to find some time to pull together another collection of links and articles pertaining to various data center technologies. Feedback is always welcome!

Networking

  • VMware has released a PowerCLI preview/fling for NSX-T; Kyle Ruddy has a write-up here. Looks like this preview provides some high-level cmdlets for NSX-T that weren’t available before.
  • Cilium, the open source project working to bring eBPF-powered networking and security to Kubernetes environments, has hit the 1.0 release. I will freely admit that I am a fan of what the Cilium folks are doing.
  • What’s that? Don’t know what eBPF is? Or XDP? Not to worry, the nice folks over at Netronome have a post that explains it all.
  • People do all kinds of interesting things with Raspberry Pis; here’s an article by Scott Helme on using a Pi to secure DNS traffic.

Servers/Hardware

  • This is more of a follow-up to one of my own articles than a pointer to someone else’s article. After continued use (including on business trips) of my Lenovo ThinkPad X1 Carbon running Fedora 27, I continue to be impressed Continue reading

Avoid Write-Only Code

You probably know that fantastic feeling when you think your newly-discovered tool is a Hammer of Thor, capable of solving every problem (or at least crashing through it). I guess you’re also familiar with that sinking feeling when you’re trying to use your beloved hammer to whitewash a bikeshed.

Not surprisingly, the cruder the tool is, the quicker you’ll hit its limits, like when you try to do data processing in Jinja2 (hint: don’t).

Read more ...

SmoothOperator: reducing power fragmentation and improving power utilization in large-scale datacenters

SmoothOperator: reducing power fragmentation and improving power utilization in large-scale datacenters Hsu et al., ASPLOS’18

What do you do when your theory of constraints analysis reveals that power has become your major limiting factor? That is, you can’t add more servers to your existing datacenter(s) without blowing your power budget, and you don’t want to build a new datacenter just for that? In this paper, Hsu et al. analyse power utilisation in Facebook datacenters and find that overall power budget utilisation can be comparatively low, even while peak requirements are at capacity. We can’t easily smooth the workload (that’s driven by business and end-user requirements), but maybe we can do something to smooth the power usage.

Our experiments based on real production workload and power traces show that we are able to host up to 13% more machines in production, without changing the underlying power infrastructure. Utilizing the unleashed power headroom with dynamic reshaping, we achieve up to an estimated total of 15% and 11% throughput improvement for latency-critical service and batch service respectively at the same time, with up to 44% of energy slack reduction.

No more headroom and low utilisation…

There’s a maximum safe amount of power Continue reading

The Slow But Sure Return Of AMD In The Datacenter

It has been more than a decade since AMD was a force in computing in the datacenter. For that reason, we have not wasted a lot of time going over the ins and outs of its quarterly financials. But now that the Epyc CPUs and Radeon Instinct GPU accelerators are getting traction among hyperscalers, cloud builders, and selected enterprises, it is time to start keeping an eye on how AMD is doing financially.

With most of the financial analysis that we do here at The Next Platform, we use the middle of the Great Recession, in the first quarter …

The Slow But Sure Return Of AMD In The Datacenter was written by Timothy Prickett Morgan at The Next Platform.

IDG Contributor Network: How to achieve and sustain peak performance in your data center

As Tom Brady once said, “if you don’t play to win, don’t play at all.” This mantra is reflected in Brady’s TB12 Method, a holistic lifecycle based on 12 principles that make up the optimal approach to exercise, training and living a life of vitality. It’s built on the premise of stopping an accident prior to it happening through pre-habitation and taking intelligent, strategic preventative measures. And while implementing all 12 principles is not required, their effect is cumulative: the more you can incorporate, the better your results will be.While this methodology applies to training superior athletes, data center managers have their own set of principles to ensure optimal data center performance. What we’ve dubbed the DCM10 Method, there are 10 fundamental steps every data center manager should use to evaluate their current data center and help transform its functionality. Similar to the TB12 Method, while not all of these steps are required to create a successful strategy, the more best practices applied will result in higher performance across the board.To read this article in full, please click here

IDG Contributor Network: How to achieve and sustain peak performance in your data center

As Tom Brady once said, “if you don’t play to win, don’t play at all.” This mantra is reflected in Brady’s TB12 Method, a holistic lifecycle based on 12 principles that make up the optimal approach to exercise, training and living a life of vitality. It’s built on the premise of stopping an accident prior to it happening through pre-habitation and taking intelligent, strategic preventative measures. And while implementing all 12 principles is not required, their effect is cumulative: the more you can incorporate, the better your results will be.While this methodology applies to training superior athletes, data center managers have their own set of principles to ensure optimal data center performance. What we’ve dubbed the DCM10 Method, there are 10 fundamental steps every data center manager should use to evaluate their current data center and help transform its functionality. Similar to the TB12 Method, while not all of these steps are required to create a successful strategy, the more best practices applied will result in higher performance across the board.To read this article in full, please click here

Jazmin Fallas Kerr: Creating Opportunities for Women in Costa Rica

The age-old tradition of the physical marketplace may be crumbling, and many women-led families in Costa Rica are feeling the pinch. In fact, in Jazmin Fallas Kerr’s hometown, Desamparados, nearly half of all families with women as head of household are in poverty.

To combat that, Kerr made a digital bridge between creation and commerce. Hyena is an Internet-based marketplace which allows women artisans to sell their handiwork online for a fair price. The site now has more than 50 local women courting customers for their crafts. Kerr’s upbringing helped her develop the idea.

“I was inspired by my mother; she is a family leader and artisan,” Kerr said. “We are from an area of social vulnerability, and I know the difficulties these families face.”

The site helps solve a more universal problem. It gives women the flexibility and time to work outside of homemaking responsibilities.

“[I want to] empower women to create better opportunities for their families,” Kerr said. “It is being achieved through a platform involving different actors in society that involves from fair trade to online education.”

Winner of the “Creator of Digital Change” contest, Hyena exists thanks to that $3,000 prize money from the Continue reading

5 reasons to buy refurbished Cisco equipment

It’s a pretty common scenario. You’re considering implementing Cisco hardware in your infrastructure, and you talk to a vendor who pushes a shiny, new, cutting-edge solution on you.It’s impressive, this bleeding-edge solution — it has all sorts of bells and whistles (whether your business plan can make smart use of them or not) and it comes with a nice, healthy price tag.So, you’re left with a really difficult question. Should you stretch your already thin budget with a solution that may actually be overkill for your needs? Or do you have options[ Read also: 80/20 rule of network equipment: Stay on budget and innovate. | Get daily insights: Sign up for Network World newsletters. ] 5 things to consider when buying network hardware Keep these five factors in mind as you consider how to proceed with your Cisco hardware solutions because that vendor who thinks you should buy new probably doesn’t want you to know them.To read this article in full, please click here

5 reasons to buy refurbished Cisco equipment

It’s a pretty common scenario. You’re considering implementing Cisco hardware in your infrastructure, and you talk to a vendor who pushes a shiny, new, cutting-edge solution on you.It’s impressive, this bleeding-edge solution — it has all sorts of bells and whistles (whether your business plan can make smart use of them or not) and it comes with a nice, healthy price tag.So, you’re left with a really difficult question. Should you stretch your already thin budget with a solution that may actually be overkill for your needs? Or do you have options[ Read also: 80/20 rule of network equipment: Stay on budget and innovate. | Get daily insights: Sign up for Network World newsletters. ] 5 things to consider when buying network hardware Keep these five factors in mind as you consider how to proceed with your Cisco hardware solutions because that vendor who thinks you should buy new probably doesn’t want you to know them.To read this article in full, please click here

Cisco reinforces storage with new switches, mgmt. software

Cisco this week fortified its storage family with two new 32G Fibre Channel switches and software designed to help customers manage and troubleshoot their SANS.The new switches, the 48-Port MDS 9148T and 96-Port MDS 9396T feature a technology called Auto Zone that automatically detects any new  storage servers or devices that log into a SAN and automatically zones them without having to do manual configuration.[Check out REVIEW: VMware’s vSAN 6.6 and see IDC’s top 10 data center predictions . | Get regularly scheduled insights by signing up for Network World newsletters. ] The idea is to eliminate the cycles spent in provisioning new devices and avert errors that typically occur when manually configuring complex zones. Even when a host or storage hardware is upgraded or a faulty facility is replaced, the switch automatically detects the change and zones them into the SAN, Cisco said.To read this article in full, please click here

Cisco reinforces storage with new switches, mgmt. software

Cisco this week fortified its storage family with two new 32G Fibre Channel switches and software designed to help customers manage and troubleshoot their SANS.The new switches, the 48-Port MDS 9148T and 96-Port MDS 9396T feature a technology called Auto Zone that automatically detects any new  storage servers or devices that log into a SAN and automatically zones them without having to do manual configuration.[Check out REVIEW: VMware’s vSAN 6.6 and see IDC’s top 10 data center predictions . | Get regularly scheduled insights by signing up for Network World newsletters. ] The idea is to eliminate the cycles spent in provisioning new devices and avert errors that typically occur when manually configuring complex zones. Even when a host or storage hardware is upgraded or a faulty facility is replaced, the switch automatically detects the change and zones them into the SAN, Cisco said.To read this article in full, please click here

1 … 1,584 1,585 1,586 1,587 1,588 … 3,795