Clare Hopping

Clare Hopping

As you know, enterprise IT conversation over the past year have often centered upon the open-source Kubernetes container orchestration system. In fact, Kubernetes has emerged as the key technology — and even primary platform — of cloud migrations for a wide variety of organizations.

Kubernetes is critical to forward-looking enterprises that continue to push their IT infrastructures toward maximum functionality, scalability, and flexibility.

read more

Technology has changed tremendously in the last 20 years. From onion architectures to APIs to microservices to cloud and containers, the technology artifacts shipped by teams has changed. And that’s not all – roles have changed too. Functional silos have been replaced by cross-functional teams, the skill sets people need to have has been redefined and the tools and approaches for how software is developed and delivered has transformed. When we move from highly defined rigid roles and systems to more fluid ones, we gain agility at the cost of control. But where do we want to keep control? How do we take advantage of all these new changes without losing the ability to efficiently develop and ship great software? And how should program and project managers adapt?

read more

Amazon Web Services (AWS) will release its latest GPU-equipped instance with support for NVIDIA’s T4 Tensor Core GPUs with a particular focus on machine learning workloads.

The specs, while not fully there yet, promise so far AWS-custom Intel CPUs, up to 384 gibibytes (GiB) of memory, up to 1.8 TB of fast local NVMe storage, and up to 100 Gbps in networking capacity.

From NVIDIA’s side, T4 will be supported by Amazon Elastic Container Service for Kubernetes. “Because T4 GPUs are extremely efficient for AI inference, they are well-suited for companies that seek powerful, cost-efficient cloud solutions for deploying machine learning models into production,” Ian Buck, NVIDIA general manager and vice president of accelerated computing wrote in a blog post.

“NVIDIA and AWS have worked together for a long time to help customers run compute-intensive AI workloads in the cloud and create incredible new AI solutions,” added Matt Garman, AWS vice president of compute services in a statement. “With our new T4-based G4 instances, we’re making it even easier and more cost-effective for customers to accelerate their machine learning inference and graphics-intensive applications.”

The announcement came at NVIDIA’s GPU Technology Conference in San Jose alongside various other news. From the AWS stable it was also announced that the NVIDIA Jetson AI platform now supports robotics service AWS RoboMaker, while AI Playground, as reported by sister publication AI News, is an accessible online space for users to become familiar with deep learning experiences.

It’s worth noting that AWS is not the first company to score in this area. In November Google touted itself as the ‘first and only’ major cloud vendor to offer T4 compatibility. In January, the company announced its T4 GPU instances were available in beta across six countries.

Interested in hearing industry leaders discuss subjects like this and sharing their experiences and use-cases? Attend the Cyber Security & Cloud Expo World Series with upcoming events in Silicon Valley, London and Amsterdam to learn more.

We are witnessing significant uplift in data transmission speeds offered by network connectivity providers. Service providers are now promising speeds in hundreds of MBs to GBs with which, for instance, we can see live streaming of Blu-ray movie prints without any buffering.

Such network speeds are set to trigger many new technology possibilities. Businesses cannot afford to stay behind, as they have to take into account new technologies which are widely adopted by the competitive market landscape. Therefore, the focus of businesses has become clear and narrow; to constantly satisfy customer demands with lucrative digital offerings and push businesses ahead for gaining competitive advantage.

To align with this trend, businesses have already started to optimise and redesign their data centres to handle a vast amount of data generated by a growing number of consumer devices. It is obvious for businesses to transform the data centre for addressing the need for upgrading. A transition would involve the use of:

• Virtual network functions (VNFS), which replaces server hardware with software-based packages to specific work – network function virtualisation (NFV)
• Software-defined networking to gain a central control of the network using a core framework which will allow admins to define network operations and security policies
• Seamless orchestration among several network components using ONAP, ETSI OSM, Cloudify among others
• Workloads (VM and containers) and data centre management by implementing OpenStack, Azure Stack, Amazon S3, CloudStack, and Kubernetes. Containers are getting widely adopted due to features like faster instantiation, integration, scaling, security, and ease in management

The next thing which will disrupt the data centre is the adoption of edge architecture. Edge computing will bring a mini data centre closer to where data is going generated by devices like smartphones, industrial instruments, and other IoT devices. This will add more endpoints before data is gathered by the central data centre. But the advantage is that maximum computing will be done at the edge that will help to reduce the load on network transmission resources. Adding to this, hyperconvergence can be used at edge nodes to bring simplification in the required mini data centre.

Mobile edge computing (MEC), a core project maintained by ETSI, is emerged at an edge computing model to be followed by telecom operators. ETSI is maintaining and working on innovations to improve the delivery of core network functionalities using MEC, as well as guiding vendors and service providers.

Aside from edge computing, network slicing is a new architecture introduced in 5G that will have an impact on how data centres are designed for particular premises, and dedicated for specific cases such as Industrial IoT, transportation, and sports stadia.

Data centre performance for high speed networks

In this transforming age, a large amount of data will transfer between devices and the data centre as well as between data centres. As low latency and high bandwidth is required by new use cases, it is important to obtain a higher level of performance from the data centre. It is not possible to achieve such paramount performance with legacy techniques and by adding more capacity to data centres.

With the ‘data tsunami’ of recent years, data centre technology vendors came up with new inventions and communities formed to address performance issues raised by different types of workloads. One of the techniques which has been significantly utilised in new age data centres is to offload some of the CPU tasks to network or server interconnecting switches and routers. Let’s take an example of the network interface card (NIC) which, when used to connect servers to network components of the data centre, has become a SmartNIC, offloading processing tasks that the system CPU would normally handle. SmartNICs can perform network-intensive functions such as encryption/decryption, firewall, TCP/IP, and HTTP processing.

Analyst firm Futorium conducted a Data Centre Network Efficiency survey targeted to IT professionals about their perceptions and views on data centres and networks. Apart from virtualising network resources and workloads, for efficient processing of data for high-speed networks, SmartNIC usage and process offload techniques have emerged as the top interest for IT professionals. This reveals how businesses are relying more on smart techniques which can save costs, along with notable data centre performance improvements for faster networks.

Workload accelerators, like GPUs, FPGAs, and SmartNICs are widely used in current enterprise and hyperscale data centres to improve data processing performance. These accelerators interconnect with CPUs for generating faster processing of data and require much lower latency for transmitting data back and forth from the CPU server.

Most recently, to address the high speed and lower latency requirements between workload accelerators and CPUs, Intel, along with companies including Alibaba, Dell EMC, Cisco, Facebook, Google, HPE and Huawei, have formed an interconnect technology called Compute Express Link (CXL) that will aim to improve performance and remove the bottlenecks in computation-intensive workloads for CPUs and purpose-built accelerators. CXL is focused to create high speed, low latency interconnect between the CPU and workload accelerators, as well as maintain memory coherency between the CPU memory space and memory on attached devices. This allows for resource sharing for higher performance, reduced software stack complexity, and lower overall system cost.

NVMe is another interface introduced by the NVM Express community. It is a storage interface protocol used to boost access to SSDs in a server. NVMe can minimise CPU cycles from applications and handle enormous workloads with lesser infrastructure footprints. NVMe has emerged as a key storage technology and has had a great impact on businesses, which are dealing with vast amounts of fast data particularly generated by real-time analytics and emerging applications.

Automation and AI

Agile 5G networks will result in the growth of edge compute nodes in network architecture to process data closer to endpoints. These edge nodes, or mini data centres, will sync up with a central data centre as well as be interconnected to each other.

For operators, it will be a task ahead to manually set up several edge nodes. The edge nodes will regularly need initial deployment, configuration, software maintenance and upgrades. In the case of network slicing, there could be a need to install, or update VNFs for particular tasks for devices in the slice. It is not possible to do this manually. At this point, automation comes into the picture where operators need to get the central dashboard at the data centre to design and deploy configuration for edge nodes.

Technology businesses are demonstrating or implementing AI and machine learning at the application level for enabling auto-responsiveness – for instance, using chatbots on a website. Much of the AI is applied for a data lake, to generate insights from self-learning AI-based systems. These types of autonomous capabilities will be required by the data centre.

AI systems will be used for monitoring server operations for tracking activities meant for self-scaling for a sudden demand in compute or storage capacity, as well as self-healing from breakdowns, and end-to-end testing of operations. Already, tech businesses have started offering solutions for each of these use cases; for example, a joint AI-based integrated infrastructure offering by Dell EMC Isilon and NVIDIA DGX-1 for self-scaling at the data centre level.

Conclusion

New architectures and technologies are being introduced with the revolution in the network. Most of this infrastructure has turned software-centric as a response to the growing number of devices and higher bandwidth. Providing lower latency – up to 10 microseconds – is a new challenge for operators to enable new technologies in the market. For this to happen, data centres need to complement the higher broadband network. It will form the base for further digital innovation to occur.

Editor’s note: Download the eBook ‘5G Architecture: Convergence of NFV & SDN Networking Technologies’ to learn more about the technologies behind 5G and the status of adoption, along with key insights into the market

The post Analysis: How are Faster Networks Advancing the New-Age Datacenters appeared first on Calsoft Inc. Blog.

Interested in hearing industry leaders discuss subjects like this and sharing their experiences and use-cases? Attend the Cyber Security & Cloud Expo World Series with upcoming events in Silicon Valley, London and Amsterdam to learn more.

Clare Hopping

Last week, Amazon Web Services (AWS) announced it was launching an open source value-added distribution for search and analytics engine Elasticsearch. As AWS evangelist Jeff Barr put it, the launch would “help continue to accelerate open source Elasticsearch innovation” with the company “strong believers in and supporters of open source software.”

Yet for industry-watchers and those sympathetic to the open source space, this has been seen as the latest move in a long-running spat between the developers and software vendors on one side, and the cloud behemoths – in particular AWS – on the other. So who is right?

Previous moves in the market have seen a lot of heat thrown in AWS’ direction for, as the open source vendors see it, taking open source code to which they have not contributed and selling software as a service around it. MongoDB, Confluent and Redis Labs were the highest profile companies who changed their licensing to counter this threat, with reactions ranging from understanding through gritted teeth to outright hostility.

In December, Confluent co-founder Jay Kreps outlined the rationale for changing licensing conditions. “The major cloud providers all differ in how they approach open source,” he wrote in a blog post. “Some of these companies partner with the open source companies that offer hosted versions of their system as a service. Others take the open source code, bake it into the cloud offering, and put all their own investments into differentiated proprietary offerings.

“The point is not to moralise about this behaviour – these companies are simply following their commercial interests and acting within the bounds of what the license of the software allows,” Kreps added. “We think the right way to build fundamental infrastructure layers is with open code. As workloads move to the cloud we need a mechanism for preserving that freedom while also enabling a cycle of investment. This is our motivation for the licensing change.”

Redis Labs, when changing its licensing stipulations for the second time last month after developers voiced their concerns, sounded a note of cautious optimism. The company had noted how it was ‘seeing some cloud providers think differently about how they can collaborate with open source vendors.’ Speaking to CloudTech at the time, CEO Ofer Bengal said, AWS aside, “the mood [was] trying to change.”

So whither Amazon’s announcement last week? Several pundits have noted that AWS potentially can’t win in these situations; if the company doesn’t contribute to a particular project it is stripping the technology away, but if it does it is impacting competitors.

AWS VP cloud architecture strategy Adrian Cockcroft – who said on Twitter the company had “proposed to give back jointly at a significant level and [was] turned down” – noted its official stance. Cockcroft had seemingly given short shrift to the moves Redis et al were making, saying there were examples where maintainers were ‘muddying the waters between the open source community and the proprietary code they create to monetise the open source.’

The logical response for a cloud supplier to the addition of adverse licensing terms will be in some cases a fork. The question of blame is difficult for non-partisans to assign – both parties are essentially acting as might be expected

“At AWS, we believe that maintainers of an open source project have a responsibility to ensure that the primary open source distribution remains open and free of proprietary code so that the community can build on the project freely, and the distribution does not advantage any one company over another,” he wrote.

“When the core open source software is completely open for anyone to use and contribute to, the maintainer (and anyone else) can and should be able to build proprietary software to generate revenue,” Cockcroft added. “However, it should be kept separate from the open source distribution in order not to confuse downstream users, to maintain the ability for anyone to innovate on top of the open source project, and to not create ambiguity in the licensing of the software or restrict access to specific classes of users.”

Shay Banon, CEO of Elastic, does not see it the same way. A day after AWS announced Open Distro for Elasticsearch, Banon wrote a missive with three constant themes; keeping things open, not being distracted by overtures from elsewhere, and maintaining a stellar experience for users. “Our commercial code has been an ‘inspiration’ for others,” Banon wrote. “It has been bluntly copied by various companies, and even found its way back to certain distributions or forks, like the freshly minted Amazon one, sadly, painfully, with critical bugs.

“When companies came to us, seeing our success, and asked for [a] special working relationship in order to collaborate on code, demanding preferential treatment that would place them above our users, we told them no,” Banon added. “This happened numerous times over the years, and only recently again, this time with Amazon. Some have aligned and became wonderful partners to us and the community. Others, sadly, didn’t follow through.”

So what happens from here? Can all parties see eye to eye? Stephen O’Grady, principal analyst at RedMonk and a long-time follower of the space, noted both sides of the argument. “The Amazon and Elastic controversy is the product of a collision of models,” O’Grady wrote. “To Banon and Elastic’s credit, Elasticsearch proved to be an enormously popular piece of software. The permissive license the software was released under enabled that popularity… [but] permissive licenses also enable usage such as Amazon’s.

“The logical response for a cloud supplier to the addition of adverse licensing terms will be, in at least some cases, a fork, which is why a move like Amaozn’s… was expected and inevitable,” O’Grady added. “It is also why the question of blame is difficult for non-partisans to assign. Both parties are essentially acting as might be expected given their respective outlooks, capabilities and legal rights.”

Ultimately, this is one which will run and run. As O’Grady explained, the status quo is likely to persist. “It is probable that Amazon’s move here will be the first but not the last,” he added. “As it and other cloud providers attempt to reconcile customer demand with the lack of legal restrictions against creating projects like the Open Distro for Elasticsearch, they are likely to come to the inescapable conclusion that their business interests lie in doing so.

“The incentives and motivations of both parties are clear, and understandable and logical within the context of their respective models,” he added. “Models which are and will continue to be intrinsically at odds even as they’re inextricably linked.”

Interested in hearing industry leaders discuss subjects like this and sharing their experiences and use-cases? Attend the Cyber Security & Cloud Expo World Series with upcoming events in Silicon Valley, London and Amsterdam to learn more.

Clare Hopping