NetworkingNexus.net

netdev 0x12 Update on Software Gone Wild

In recent years Linux networking started evolving at an amazing pace. You can hear about all the cool new stuff at netdev conference… or listen to Episode 94 of Software Gone Wild to get a CliffsNotes version.

Roopa Prabhu, Jamal Hadi Salim, and Tom Herbert joined Nick Buraglio and myself and we couldn’t help diverging into the beauties of tc, and the intricacies of low-latency forwarding before coming back on track and started discussing cool stuff like:

RobinHood: tail latency aware caching – dynamic reallocation from cache-rich to cache-poor

RobinHood: tail latency aware caching – dynamic reallocation from cache-rich to cache-poor Berger et al., OSDI’18

It’s time to rethink everything you thought you knew about caching! My mental model goes something like this: we have a set of items that probably follow a power-law of popularity.

We have a certain finite cache capacity, and we use it to cache the most frequently requested items, speeding up request processing.

Now, there’s a long tail of less frequently requested items, and if we request one of these that’s not in the cache the request is going to take longer (higher latency). But it makes no sense whatsoever to try and improve the latency for these requests by ‘shifting our cache to the right.’

Hence the received wisdom that unless the full working set fits entirely in the cache, then a caching layer doesn’t address tail latency.

So far we’ve been talking about one uniform cache. But in a typical web application one incoming request might fan out to many back-end service requests processed in parallel. The OneRF page rendering framework at Microsoft (which serves msn.com, microsoft.com and xbox.com among others) relies on more than 20 backend Continue reading

Mesosphere Revs Up DC/OS With a Kubernetes Engine

The Mesosphere Kubernetes Engine allows for multi-cloud and infrastructure management from a single pane of glass and reduces cluster sprawl.

Cloud Business Sends Intel Data Center Revenue Soaring

The chip giant’s third-quarter fiscal 2018 revenue of $19.2 billion was an all-time record, up 19 percent year over year, with all business groups posting record revenue.

The Top 6 Questions You Asked on Containerizing Production Apps

We recently hosted IDC research manager Gary Chen as a guest speaker on a webinar where he shared results from a recent IDC survey on container and container platform adoption in the enterprise. IDC data shows that more organizations are deploying applications into production using containers, driving the need for container platforms like Docker Enterprise that integrate broad management capabilities including orchestration, security and access controls.

The audience asked a lot of great questions about both the IDC data and containerizing production applications. We picked the top questions from the webinar and recapped them here.

If you missed the webinar, you can watch the webinar on-demand here.

BrandPost: Top Ten Reasons to Think Outside the Router – No. 10: It’s Getting Cloudy

Borrowing from the iconic David Letterman Top Ten List segment from his former Late Show, this new blog series will countdown the Top Ten Reasons to Think Outside the Router.The #10 reason it’s time to retire traditional routers at the branch: It’s getting cloudy! In fact, it’s already cloudy. In November 2017, Forrester projected that 2018 would be the year that more than 50% of enterprise applications would be hosted in public and private clouds. Here we are in 2018, and 96% of 997 SMB and enterprise companies surveyed now use cloud services. The migration to cloud-based applications and infrastructure continues to accelerate and is happening faster than anyone predicted. The challenge: enterprise router-centric WAN architectures weren’t designed for the cloud.To read this article in full, please click here

BT Replaces Departing Chief Executive With Worldpay’s Philip Jansen

The London-based telecom provider’s current Chief Executive Gavin Patterson announced in June that he would leave BT amid company struggles.

Rackspace launches disaster recovery as a service program

Give managed cloud computing provider Rackspace points for timing. Coming right after the Uptime Institute issued a warning for data center operators to improve their environmental disaster plans, the company announced it is broadening its existing disaster recovery as a service (DRaaS) program for on-premises, colocation, and multi-cloud environments.The expansion utilizes Zerto’s disaster recovery software, which is specifically designed to provide business continuity and disaster recovery in a cloud and virtualized environment.To read this article in full, please click here

Rackspace launches disaster recovery as a service program

Worth Reading: Spanning clouds with Kubernetes

The use of Kubernetes and other opens standards also means that enterprises can place workloads not only on IBM’s private or public clouds, but into environments of other cloud providers that conform to the container orchestration platform. —Jeffrey Burt @The Next Platform

AMD Inches Up Server Share On The Way To Rome

IT shops are notorious when it comes to avoiding risk, and this is something that we should be grateful for because it increases the odds that this increasingly compute-dependent world keeps working. …

AMD Inches Up Server Share On The Way To Rome was written by Timothy Prickett Morgan at .

Gartner Ranks VMware Highest in Its Inaugural WAN Edge Magic Quadrant

VMware has more than 2,000 SD-WAN customers, while the second and third ranked SD-WAN vendors count about 1,000 customers, each.

Trump Wants a National 5G Spectrum Policy

The president issued a memorandum that is intended to make more spectrum available so that operators can more quickly deploy 5G.

OpenStack Foundation releases software platform for edge computing

The OpenStack Foundation, the joint project created by NASA and Rackspace to create a freely usable Infrastructure-as-a-Service (IaaS) platform, has announced the initial release of StarlingX, a platform for edge computing.StarlingX is designed for remote edge environments, offering node configuration in host, service management, and perform software updates remotely. It can also warn operators if there are any issues with the servers or the network.The foundation says the platform is optimized for low-latency, high-performance applications in edge network scenarios and is primarily aimed at carrier networking, industrial Internet of Things (IIoT), and Internet of Things (IoT).To read this article in full, please click here