Connect with us


An open-source mobile network project • The Register

Voice Of EU



Systems Approach This month’s column was co-written by Amar Padmanabhan, a lead developer of Magma, the open-source project for creating carrier-grade networks; and Bruce Davie, a member of the project’s technical advisory committee.

Discussions about mobile and wireless networking seem to attract buzzwords, especially with the transition to 5G. And hence we see a flurry of “cloudification” of mobile networking equipment—think containers, microservices, and control and user plane separation.

But making an architecture cloud native is more than just an application of buzzwords: it entails a number of principles related to scale, tolerance of failure, and operational models. And indeed it doesn’t really matter what you call the architecture; what matters is how well it works in production.

In this post we have tried to articulate some of the defining traits that have guided the development of Magma, in pursuit of a cost-effective and easy-to-operate solution for mobile networks in less well-developed areas.

Commodity hardware

Most traditional networking equipment is proprietary, bundling the software with precisely specified and configured hardware. But Magma, like most cloud native systems, leverages low-cost, commodity hardware. Performance is achieved by the use of scale-out approaches, and reliability is achieved by software techniques that deal with failures of unreliable hardware. Scale-out and planning for failure are themselves key tenets of cloud native architecture, as discussed below.

From its inception, Magma has been designed to be easy to operate in diverse environments on commodity hardware. Any component can be replaced with minimal cost and disruption to the network.

Scale-out rather than scale-up

Cloud native systems typically achieve scale by horizontally adding more commodity devices, rather than scaling up the capacity of individual, monolithic systems. Magma is based on a distributed architecture that scales out. Capacity is increased by adding small devices throughout the network — for example, networks can deploy hundreds of access gateways around radio towers instead of adding a few large boxes to the core as is common in traditional EPC (evolved packet core) designs. This distributed architecture is also important when it comes to our next point, designing for failure.

Small fault domains

In any cloud system, it is expected that individual components will fail, and failure is treated as a common part of the operational flow of the system. Many of Magma’s design decisions flow from that premise. In traditional telco architectures, by contrast, failures are assumed to be rare and are handled through specific exception paths, like hot standbys and fully redundant services.

A failure should impact as few users as possible (i.e., fault domains must be small) and should not affect other components. For example, a failure in a small access gateway might affect only a few hundred customers. Conversely, if a network built upon two large cores has a failure in one of its cores, half of its customers may lose service.

It’s not sufficient to just divide a large monolith into smaller components. You also need to localize the state within components to limit the impact of failures. Magma does this by localizing the state associated with any given user equipment (UE) to a single access gateway. Thus, the impact of component failure is limited: only the UEs served by a given access gateway are impacted. The access gateway is the location for per-UE “runtime state,” which depends on events such as the powering on of UEs or movement of a UE to the coverage area of a new base station. By contrast, UE runtime state tends to be spread among components in traditional 3GPP implementations.

While runtime state is localized to the relevant access gateway, configuration state is stored centrally in the Magma orchestrator, because configuration is a network-wide property that is provided through the central API. If a component of the orchestrator fails, it only prevents updates to configuration, but does not impact the runtime state; hence, UEs can continue to operate even if the orchestrator is restarting.

Simplified operations

The scalability of cloud native systems applies as much to operations as to performance. Centralized control planes, such as those found in software-defined networks (SDN), have emerged as a way to simplify the operations of networks. Indeed, Magma was influenced by the experience of building Nicira’s SDN system. Whereas centralized control was once viewed as an unacceptable single point of failure or a scaling bottleneck, it is now well understood that reliable, logically centralized controllers can be built from a collection of commodity servers. It’s much simpler to operate a whole network when it is viewed from a central point of control, rather than determining how to configure each network device individually.

Diagram showing the Magma cloud native architecture

Magma’s architecture of distributed access gateways and logically centralized control

The logically centralized point of control in Magma is the orchestrator, and it loosely corresponds to a controller in an SDN system. It is implemented on a set of machines (typically three), any one of which may fail without bringing down the orchestrator. This set of machines exposes a single API from which the network as a whole may be configured and monitored. As the size of the network increases and more access gateways are added, the operator maintains a single, centralized view of the network.

Access gateways represent a distributed data plane implementation and also contain local control-plane components. The federation gateway implements a set of standard protocol interfaces to allow a Magma-enabled system to interoperate with standard cellular networks.

Like SDN systems, Magma separates the control plane and data plane. This is essential to the simplified operational model, but also affects reliability. Failure of a control-plane element does not cause the data plane to fail, although it may prevent a new data-plane state from being created (e.g., bringing new UE online). This is a more complete separation than provided by the CUPS specification of 3GPP as we’ve discussed previously.

Magma’s data plane, running in the access gateways, is software-controlled and programmed through well-defined and stable interfaces (similar in principle to OpenFlow) that are independent of hardware. Again, this is the same approach as taken in SDN and enables the data plane to leverage commodity hardware and easily evolve over time.

Desired state model

Similar to many cloud native systems (like Kubernetes), Magma uses a desired state configuration model. APIs allow users or other software systems to configure the intended state, while the control plane is responsible for ensuring that this state is realized. The control plane takes on the responsibility of mapping from the operator’s intended state to the actual implementation in the access gateways, reducing the operational complexity of managing a large network.

The desired state model has been proven effective in other cloud native contexts because it enables components to simply compare the current state with the desired state, and then make adjustments as needed. For example, if the desired state is that there are two sessions active, the control plane monitors the system to ensure it meets the requirement and takes action to activate a session if it is missing. (Here is Joe Beda with my favourite description of how desired state works in Kubernetes.)

By contrast, traditional 3GPP implementations have used a “CRUD” (create, read, update, delete) model, in which a sequence of actions (create new session or update session, for example) determine the state. In the event of a message loss or component failure, it becomes hard to determine if the current state of a component is correct.

While many of these design decisions might seem obvious, they are fairly different from the principles of the standard 3GPP architecture. For example, even though 3GPP has the concept of CUPS, control plane elements often hold some user plane (data plane) state. A proper separation of control and data planes leads to a more robust architecture with better upgradability.

Ultimately, it doesn’t matter what we call the architecture. What matters is how well it scales, how it handles failures, and how easy it is to operate. By adopting technologies and principles from cloud services, Magma delivers a mobile network solution that is robust, leverages commodity hardware, scales from small to large deployments gracefully, and offers a central point of control for operational simplicity. We’re gathering operational data as I write this; our goal is to quantify these claims in a later piece. ®

Source link


Angharad Yeo: the 10 funniest things I have ever seen (on the internet) | Comedy

Voice Of EU



I am a child of the internet. I was always drawn to computers and tech, and used to beg my dad to bring us to his office on a weekend so we could use the high-speed internet to play Neopets games. As I got older it was all MSN, MySpace, Paramore fan forums, Tumblr, Twitter and now TikTok. I want nothing more than to zone out and look at my little pictures.

One of my favourite things about the internet is that it allows you to see everyone’s best joke. The moment in their life where they were at their absolute funniest – whether it be because they had a moment of brilliant wit or because they got pulled through a panel roof while practising for a high school play (I assume).

The internet has rotted my brain with the following content. Please now allow it to rot yours.

Allow TikTok content?

This article includes content provided by TikTok. We ask for your permission before anything is loaded, as they may be using cookies and other technologies. To view this content, click ‘Allow and continue’.

The Pandemic Years have (and continue to be) difficult for everyone. Who among us has not, at one time or another, needed to just explain themselves by saying: “It’s mental illness, innit?”

2. Perfect burger

When I showed this video to my fiancee, she flatly said: “I like how absurdist it is.” That’s her code for, “I don’t get it, but I’m happy you’re happy.” And I am happy. Look at how confident and brave this burger is – ready to take on the world, come what may. I wish to be the burger.

Allow TikTok content?

This article includes content provided by TikTok. We ask for your permission before anything is loaded, as they may be using cookies and other technologies. To view this content, click ‘Allow and continue’.

I have been to court precisely once because I inadvertently got in a cop’s way and he was grumpy about it so he booked me. The penalty was dismissed but not before I cried in front of the judge trying to explain what happened because I was so stressed out. Court is a daunting place and I simply cannot imagine walking in there with any level of irreverence. However, I’m extremely glad there are people who simply do not care, will say whatever damn thing and then an internet angel turns them into TikToks.

4. Turtle choir

This tweet is made all the more majestic by the vaguely threatening Sylvanian Families-style profile picture, on a Twitter account named @bigfatmoosepssy.

5. Trying coffee with pasta water

Climate change is slowly turning the Earth into a barren ball of pain as Mother Nature smacks us for being extremely bad. Even though individual responsibility for climate change isn’t enough to turn the tide, I still applaud those who try. Twitter user @madibskatin woke up in the morning and decided to be the change she wants to see in the world, tastebuds be damned. One could argue that it’s pretty obvious that pasta water isn’t going to make a good coffee but like my dad says as he puts pineapple juice in his coffee: “If no one tries it, how will we know? What if it’s secretly good?”

6. Soaring, flying

If you look closely, this video is actually a metaphor for the ways in which we attempt to break free from our circumstances, yet are entirely at the mercy of them.

7. You cannot trick me

This may be a parody Twitter account, but the spirit of Gail Walden speaks truths. There is no victory sweeter than that which is gained on thine enemy’s own soil.

8. Self-deprecating jokes

Humour is a coping mechanism. I am coping.

Allow TikTok content?

This article includes content provided by TikTok. We ask for your permission before anything is loaded, as they may be using cookies and other technologies. To view this content, click ‘Allow and continue’.

Dairy products are delicious. Ice-cream? Revolutionary. Cheese? Life-changing. Whipped cream on a pavlova? Essential. But milk? Disgusting. It’s not a drink, it’s a stepping stone to greater things.

Allow TikTok content?

This article includes content provided by TikTok. We ask for your permission before anything is loaded, as they may be using cookies and other technologies. To view this content, click ‘Allow and continue’.

I am absolutely 100% not at all lactose intolerant (I promise) so I don’t relate to this video at all (not even a bit).

Angharad Yeo is the host of Double J Weekends, 9am – midday, Saturdays and Sundays.

Source link

Continue Reading


F5 cuts revenue 2022 forecasts amid low network chip stocks • The Register

Voice Of EU



The artist formerly know as F5 Networks – it moved to plain old F5 in November – is clipping revenue forecasts for fiscal ’22 by $30m to $90m because it can’t source enough specialised chips to produce systems.

The continued impact of the shortfall was outlined in F5’s Q1 results to 31 December and subsequent earnings conference call, during which chief exec François Locoh-Donou opened up on the challenge of suppliers cancelling orders because they can’t meet demand.

“As a result of persistent strong system demand, our systems backlog continued to grow in Q1,” he said. “Over the last 30 days, suppliers of critical components that span a number of our platforms have informed us of significant increases in decommits.

“These came in the form of both order delivery delays and sudden and pronounced reduction in shipment quantities. The step function decline in components availability is significantly restricting our ability to meet our customers’ continued strong demand for our systems.

“Like others in the industry, we are seeing worsening availability of specialized networking chipsets. Within the last 30 days, we have learned that deliveries for 52-week lead time components or at a year ago have been pushed out and that our expected quantities have been reduced.”

Group turnover grew 10 per cent year-on-year to $687m in F5’s Q1, fuelled by a 47 per cent leap in software to $163m, 2 per cent in services to $344m, and 1 per cent in hardware to $180m.

“Our software transition continues to gain momentum,” said Locoh-Donou, adding later in the earnings call: “While we are solely disappointed that supply chain challenges have gated our ability to fulfil customer demand for systems in the near term, we are more confident than ever in our position, our strategy and our long-term opportunity.”

The backlog grew by 10 per cent so the sales pipeline is looking healthy, said the exec, who was at great pains throughout the call to tell analysts: “It absolutely is a supply issue. And the revision we’ve just done to our annual guidance is 100 per cent linked to the supply issue.”

For the year, F5 now expects sales to grow 4-8 per cent ($610m to $650m).

“The issue with our supply chain has deteriorated steadily. And last year, we were not able to ship the demand, which is why our backlog grew so much during the year.

“Things have been getting worse. And at the beginning of our fiscal year, when we were doing the planning for this year, we actually took into account the number of decommits that we were getting from various suppliers and a situation that was already very tight on a number of components.”

He said in the past month it was seeing more than 400 cancellations from suppliers, “and we were running about 30 per cent less than that even just a month ago – the situation is quite unprecedented.”

In a bid to ameliorate the supply situation, F5 said it is working to design and qualify replacement parts – which may improve thing in the second half of the year. It is also trying to pre-order more components.

F5 is confident that it will not see orders cancelled. “The demand we have is very real. Our lead times, unfortunately, have gotten progressively worse over the last five, six quarters, but we haven’t seen any increase in order cancellation, and we don’t expect to see that going forward,” Locoh-Donou stated.

Supply chain problems with silicon components have been hitting companies in the IT industry and beyond for multiple quarters now, and networking vendors are no less vulnerable.

Last year, Arista warned that lead times for key chips were extending out to 60 weeks, twice what would be expected before the pandemic. Both Arista and Juniper announced they were being forced to bump up prices in November, while Cisco warned its buyers and investors that supply chain issues were likely to persist for several months more, although it expected to see some improvement in the situation for Q3 and Q4, taking us into the second half of 2022. ®

Source link

Continue Reading


Cork data centre equipment maker Edpac acquired for €29m

Voice Of EU



Munters, a Swedish air treatment technology company, will use the Edpac acquisition to expand into the European market.

Irish data centre equipment manufacturer Edpac has been acquired by Swedish company Munters in a €29m deal.

Based in Carrigaline, Co Cork, Edpac manufactures cooling equipment and air handling systems for data centres in the European market, with additional sales in the Middle East, South America and Asia.

For Munters, which has significant operations in North America, the acquisition is an opportunity for it to expand in the European market. Once complete, the deal will see the transfer of Munters’ technologies and engineering capabilities to Ireland.

“The European data centre market is a prioritised segment for Munters, and the acquisition is a significant step in our growth strategy,” said Klas Forsström, president and chief executive of Munters.

Forsström said that Munters’ experience in the North American market will provide Edpac with “opportunities for further profitable growth” by collaborating on “technology development and establishing unified processes”.

Edpac has two manufacturing facilities in Ireland – Newmarket and Carrigaline – and employs around 150 people in the country. Currently a manufacturing partner for Munters, Edpac sees approximately 7pc of its revenue come from the sale of Munters products.

In the financial year ending April 2021, Edpac reported net sales of €17m and earnings before tax of €1.7m. According to The Irish Times, Edpac managing director Noel Lynch has led the company since it was bought from its Swiss parent in 1991.

“We are excited to welcome Edpac to Munters. Edpac brings an attractive, differentiated customer base and high-quality products,” Forsström said, adding that Edpac’s operating model “is a perfect match with Munters ways of working.”

Founded in 1955, Munters aims to create energy efficient air treatment technologies for customers in a wide range of industries. Listed on Nasdaq Stockholm, it employees 3,300 employees across 30 countries – with annual sales exceeding 7bn Swedish krona in 2020.

Don’t miss out on the knowledge you need to succeed. Sign up for the Daily Brief, Silicon Republic’s digest of need-to-know sci-tech news.

Source link

Continue Reading


Subscribe To Our Newsletter

Join our mailing list to receive the latest news and updates 
directly on your inbox.

You have Successfully Subscribed!