Monthly Archives: December 2019

Power Management Improvements Could Benefit Intel Server Performance In Linux 5.6


Some Intel server platforms could see better performance with the Linux 5.6 kernel cycle.

Intel’s Rafael Wysocki who also serves as the Linux kernel’s power management subsystem maintainer has been queuing some patches recently in working on ACPI _CST support around the Intel-Idle driver. With the final patch for using ACPI _CST on server systems with the Intel-Idle driver, Rafael explained:

In many cases, especially on server systems, it is desirable to avoid enabling C-states that have been disabled in the platform firmware (BIOS) setup, except for C1E.

As a rule, the C-states disabled this way are not listed by ACPI _CST, so if that is used by intel_idle along with the specific table of C-states that it has for the given processor, the C-states disabled through the platform firmware will not be enabled by default by intel_idle.

This updated behavior applies for server platforms back to Nehalem and each generation there after as part of the server or X-Series.

With this updated intel_idle driver behavior, Intel’s own monitoring is showing around a 12.6% improvement for FS-Mark that is a file-system benchmark but the CPU performance plays an important role too. It will be interesting to see how this Intel server power management behavioral change affects other workloads as well, which we’ll begin benchmarking shortly.

The Linux 5.6 merge window should be open around the end of January or early February. Until then the Linux power management work for Intel and other hardware will continue queuing as usual within linux-pm.git’s linux-next.

digiKam 7.0 Beta Comes With Deep Learning Powered Faces Management

The next major version of the digiKam photo management software, that is digiKam 7.0, looks promising. The digiKam Team has released their first beta with deep learning powered faces management. Since the algorithms used in background (not based on deep learning) were older and unchanged since the first revision, the team found them to be “not enough powerful” to facilitate the faces management workflow automatically.

The updated version makes use of the cascade classifier and deep neural network functionality of the OpenCV library to improve face detection big time.

[Source: TFiR]

Kubernetes Challenges for Enterprises | IT Infrastructure Advice, Discussion, Community

Kubernetes has evolved with capabilities that are suitable for enterprise IT workloads. Now that Kubernetes is ready for the enterprise, let’s learn about the challenges of implementing Kubernetes into your IT infrastructure.

Investment in Modern Technologies

In 2018, we saw a significant rise in the adoption of Kubernetes for modernization data centers by enterprises that want to leverage digital transformation. This year, we have seen more investment made by enterprise CXOs and leaders. According to the Diamanti survey for container adoption, enterprises investing $100K or more in containerization of workloads increased to 38 percent in 2019 as compared to 31 percent in 2018. But the challenge is that, despite a big hype of Kubernetes in the last 24 months, the ratio of investment in Kubernetes is quite slow. The reason for this: most leaders are assigning less IT budget, or they are still in confusion about whether to go for a single standalone framework to manage application workloads.

Deployment Types

Kubernetes can be implemented in IT infrastructure in three ways.

One is to bring on the Kubernetes framework directly into the infrastructure and re-structure and re-design the infrastructure according to architecture. The second is to hire a managed Kubernetes solution vendors to implement, deploy, and manage Kubernetes for enterprises. And the third is to turn to production solutions like Red Hat OpenShift to implement Kubernetes.

There are pros and cons to each of the solutions. Enterprise leaders must select a solution that fits their business strategy and results in an increase in service delivery and agility. The time required for this is hampering the adoption of Kubernetes in data centers. Some of the issues that must be addressed include:

DevOps: Kubernetes solves challenges to manage the container in an efficient way and provide many benefits. But Kubernetes is known to be a complex platform itself for implementation and managing DevOps aspects.

There are challenges related to automating operations like configuration of the load balancer. One can deploy an application with kubectl, but for automation, you must configure the load balancer.  

Kubernetes also is well known for the ability to efficiently utilize computing power for deployed container clusters. But to achieve this, admins have to understand that there is a manual process involved in making configurations on each pod that is hosting containers to computing resources.

Security: Kubernetes has emerged as a de facto standard tool for managing all application workloads in enterprises. Enterprises IT teams are keen to get Kubernetes in production, but on the other side, attackers are also finding new ways to breach the infrastructure powered with Kubernetes. We can say, as Kubernetes adoption has grown, so too has the attacks surface and security risks grown.

As per Dimanti container adoption survey in 2019, security has become major challenges for containers in production for enterprises.

With this, a Kubernetes platform and container security have become a major concern this year. In the last three months, Kubernetes announced major platform releases, disclosed several incidents and released several patches to fix the security holes in the Kubernetes framework. You can refer to this link to get more information.

Reason for breaches of Kubernetes based infrastructure can be:

  • Flaws in the design and architecture of Kubernetes for large or medium scales deployment.

  • If a flaw is left or overlooked by the developer in the application within the container, it can result in the compromise of containers, and further, a hacker can gain unauthorized access into clusters to make a false impact.

  • The connection of Kubernetes pods can be left unauthorized. This can result in malicious attacks on the pod itself.

  • Kubernetes infrastructure components also can be exploited, compromised, and get affected by external or internal breaches.

Such security concerns raised a new stream in the Kubernetes ecosystem in the form of securing the Kubernetes platform. Many vendors have started to offer managed solutions and services for enterprise Kubernetes platforms.

Lack of Kubernetes skills: Enterprises expect a standard and a shared platform that can be useful in multi-cloud implementation, keeping on-premises data center in place. Workloads in Kubernetes should run on any connected cloud without any problem in an agile way. For that level of output from Kubernetes, resources need to be managed by a staff with the right skills around the Kubernetes platform.

The core issue that can be raised with Kubernetes skills deficit is in performing day 2 operations for Kubernetes implementation in infrastructures like upgrades, patches, and more. Kubernetes skilled experts know how to perform backup, recovery, and applying upgrades and patches to a vast set of Kubernetes components.

Another hurdle can come up when performing a set of operations or defining a particular state from a central pane of glass. Kubernetes allows you to declare a configuration or set up instructions for the target part and set up the desired state for pods or interconnection of pods. During this process, if any error occurs, a skilled resource is required to understand the problem and the location of the error.


Kubernetes launched in 2014. In 5 years, it has slowly evolved to be adopted in mainstream IT infrastructure. As enterprises are taking Kubernetes into production, infrastructure integration and security are identified as major bottlenecks to implementation and deployment. While introducing Kubernetes as a key platform for digital transformation, enterprise leaders need to set up a good plan initially and build a properly-knowledgeable pool of resources for managing day 1 and day 2 operations. For larger Kubernetes deployment within enterprises, it is highly recommended to use an opt-out solution that will not result in wastage of the amount of time and cost.


Source link

Mesa’s Radeon R600 Gallium3D Driver Now Has NIR Support Under Review


Similar to the trend with other Mesa drivers, the Radeon R600g driver for supporting Radeon HD 2000 through Radeon HD 6000 series graphics cards has been seeing experimental work to introduce a NIR back-end for this modern intermediate representation. That R600 NIR support now has a merge request open meaning it could possibly land still for Mesa 20.0.

The R600g NIR support has been worked on by Gert Wollny and currently targets Radeon HD 5000 “Evergreen” graphics cards with support for other AMD GPU generations handled by this Gallium3D driver not yet supported. Additionally, this NIR back-end only supports vertex / fragment / geometry shaders for now and other features missing.

R600g NIR development was done so far using the Radeon HD 5450.

The preliminary support can be found via this merge request. This R600g NIR support would be disabled by default as a lot of features and optimizations are left to be tackled, but for those interested it’s one step closer to mainline for those using these old pren-GCN Radeon graphics cards.

This DIY Linux-Powered Business Card With USB Port Costs Just $3

George Hilliard, an embedded systems engineer by profession, decided to make a barebones Linux board in a business card form factor. It is a complete, minimal ARM computer that runs customized Linux firmware built with Buildroot. This Linux-powered business card has a USB port in the corner. Hilliard says that the card takes about 6 seconds to boot after plugging into a computer. It shows up over USB as a flash drive and a virtual serial port that can be used to log into the card’s shell.

[Source: Fossbytes]