Blogs Lalibre.be
Lalibre.be | Créer un Blog | Avertir le modérateur

13/12/2017

lenovo B470A Battery www.all-laptopbattery.com

All three orchestration services are designed to be scalable to any number of containers running in whatever configuration the user needs. As Docker marketing veep David Messina explained to El Reg in a briefing earlier this week, We literally mean that all these capabilities that we're talking about scale from, let's say, a two-container application that runs on my laptop, all the way up to hundreds of services running on hundreds of hosts.Public APIs are also planned for all three services, although only the API for Docker Machine is available as of Thursday. APIs for the other two are expected to launch in the first half of 2015.The debut of these new services should come as no surprise to followers of the Docker open source project, which develops the software in full public view. All three services have been available as pull requests on GitHub and they have been developed based on feedback from the Docker community. Thursday's announcement was more of a coming-out party, where Docker showed them off to more casual consumers of the tech for the first time.But not everyone agrees that piling these sorts of features onto Docker is the right approach. On Monday, CoreOS – makers of a lightweight Linux distro for massive-scale server deployments – announced Rocket, a minimalist container format created specifically in response to Docker's platform aspirations.

At a community event in San Francisco, CoreOS CEO Alex Polvi criticized Docker's process model, saying, Having a service that runs as root on your servers that does everything is fundamentally flawed. He added that with Rocket, CoreOS would focus on developing lightweight, standalone tools for managing minimal containers for large-scale web infrastructure.How this budding rivalry pans out remains to be seen. It's safe to say there will be no mass exodus from Docker any time soon, though, as the Rocket software is still pre-alpha. And the Docker ecosystem is strong and growing; partners that are already experimenting with the new orchestration services include IBM, Joyent, Microsoft, and Mesosphere, among others.Customers are now free to kick the new services around, too. More information on each is available at the Docker blog, and all three services are available in alpha versions – even including the unfinished Docker Compose – beginning on Thursday (see above links). As for when the services will be part of the shipping product, Docker says it's shooting for general availability in the second quarter of 2015. The evolution of demand changed blade systems. From that many weak but highly interchangeable origin, blades have exploded into virtually every niche. Some people needed relatively weak systems but a lot of storage. HP, among many others, delivered.

Some needed blades that could handle additional expansion cards for telecoms, instrumentation, or even GPU-based VDI. All these niches were filled.But the mainstream use case was, and remains, putting as much computing horsepower in as small a space as humanly possible while keeping the thing easy to service over the course of its life.VMEbus was created in the first place to make servicing cards in industrial systems easier. More than three decades later, ease of use remains the primary motivator behind the mainstream adoption of its descendants.What is perhaps most interesting is how the language evolved. When blade servers moved from many weak systems to beefcakes we could have collectively called them something different. We didn't.“Blades had mindshare. It was valuable from a marketing standpoint to stick with the name. Instead, we call the true inheritors of the original blade concept by new names – Moonshot, MicroCloud, fabric computing.

The meaning of the term changed with time, which is instructive when discussing the past of our industry as well as prognosticating about the future. So many of the debates we have among professionals really are down to nomenclature, and the nomenclature follows marketing, not engineering.It seems silly to think that a design that has been around in one form or another for more than 30 years will simply cease to have value. Especially when the fundamental element of blades is not so much the type of processor you cram into it but the concept of ease of use.Despite this, some people in our industry do cast aspersions on the future of blades. Popular alternatives such as the Open Compute project or Supermicro's Twin servers use unblade designs.These have a shared chassis but no shared backplane interconnecting the nodes and no chassis-wide supervisor or administrator for high-level management, configuration and monitoring.The ease of use that blades provide comes at a price. Blades are an expensive option. Hyperscale cloud providers that have to stand up tens or even hundreds of thousands of servers shy away from the cost and opt for designs that are as minimalistic and cheap as possible.But the world is not all hyperscale clouds. Lots of cloud providers are much smaller. Given the American government's propensity for doing harm to its own cloud industry, small regional clouds will be a thing for decades to come.

These providers want the blade servers’ ease of use, as do enterprises and mid-sized companies that are not ready to give up control of all their IT to the public cloud.New technologies – whether labelled cloud, unblade or anything else – will continue to emerge. This won't mean the death of blades. They will evolve into new niches.Thirty years from now we will be using their descendants and we will find them on premises, attached to pipelines and power stations, in the cloud and probably even in space. Ease of use is a hard concept to kill.Blades reflect technology. Developments tend to evolve not to supplant all those technologies that went before but to supplement them. It is very rarely a case of the survival of technology A or technology B. Instead, technology A and technology B find their place in our world.Something we should perhaps all consider the next time we are tempted to talk about technology adoption in absolutes. For Windows, the choice is Visual Studio. The Express version doesn't cost you anything and the commercial versions will set you back only a few hundred quid if you want the extra bells and whistles they give.You will want to do the job properly, and Microsoft would add that it is working to unify the environments so that your work will be transferable from mobile devices to traditional platforms.On the Mac you need the Xcode development system, which is a free download from the Apple App Store. And for Android you need the Android Developer Tools (ADT) – though in the next few months this will be superseded by the new Android Studio which is in Beta as I write.

With regard to the platform you use, you need to be running Visual Studio on Windows 8.1 if you want to develop for Windows Phone 8, and with Xcode you will need a reasonably recent flavour of Mac OS X (Snow Leopard or later).The Android developer tools are available for both Windows and Mac. They are Java-based and use the Eclipse integrated development environment (IDE), which also makes them the most open source of the lot.Having spent more years than I care to remember as a .NET developer in a Microsoft world I know my way around it pretty well already, so I tried out Xcode and ADT to see how they compare.The Android setup was the hardest to get going, primarily because it is the only one that doesn't come as a packaged, proprietary app.Since it is Java-based one has the usual fun persuading it to see the Java installation and in my case persuading it that the 64-bit version of the kit really should be able to run on the 64-bit version of Windows 7 on my HP laptop.

That said, it took only a few minutes to figure out what was wrong, and the Mac version ran with no such gremlins.Once running, the IDEs all take pretty much the same approach. The layout looks very similar: there is a pane with a hierarchical representation of all the code files and libraries included in your project; a pane with a graphical representation of the screen of the device you are developing for; and an emulator application that runs a virtual mobile device on your PC or Mac into which the IDE injects the executables so you can run and test them.You can, of course, test your code on a real device but it is a whole lot easier to run with the emulator until you are happy with the result and want to take the next step.The approach to developing applications is pretty simple: design the look and feel graphically on the GUI view of the device, then add the code that reacts to events associated with the user pressing and typing stuff.Looking at the Mac version (there have been Java and Windows developer IDEs for donkey's years), this will be refreshing for anyone who ever came across the Macintosh Programmer's Workshop, the development kit for the Mac back in the days before Mac OS X. This was slightly more complex than the user manual for a Polaris submarine. Xcode is child's play in comparison.

Les commentaires sont fermés.