Showing headlines posted by kevinlawton
Cloud Pipeline: future of inter cloud provider sneaker-nets
Shipping data quickly using intermodal transport (van, jet, boat, etc) opens up pipelining of workflows across cloud providers, with real-time marketplace-based routing to the best providers.
Server side Android, a Google version of Amazon's EC2
Server-side Android would push Android into high-value consumer/enterprise VDI spaces, and open up opportunities in server-side rendering for photo realistic gaming.
Facebook thin tablet, iTablet killer!
Forget the rumored Apple tablet. A Facebook thin tablet would rule the Earth. While the 'net is abuzz over rumors of the Apple tablet, I'd like to point out a category of device in a form-factor that doesn't yet exist, but would be a killer product. It's also what I believe the CrunchPad tablet should have been designed to be. And that's the "thin^2 tablet". By thin, I mean it's physically thin in dimension, like the iPhone, but it's also thin in the sense that thin-clients are thin when they have nothing but firmware to access a remote server.
Java performance does not scale as well as C++ with multi-tasking
Java vs C++ benchmarking rarely if ever looks at the real issue w.r.t. performance, ability to perform well within multi-tasking. In this article, multiple benchmarks are run concurrently, exposing the overhead of Java which generally at least partially slips into the cracks of extra hardware capacity on single task benchmarks. And it explains why Java is thus not a good fit for large scale computing settings, like Hadoop.
Hadoop should target C++/LLVM, not Java (because of watts)
Hadoop's Java framework will make it uncompetitive relative to Google's C++ MapReduce implementation, based on work done per watt, and will cause a lot of large-scale wasted power consumption. I believe that Hadoop should re-target C++/LLVM. Until then, Google will hold an advantage over search efficiency.
50% of Cloud Compute Power & Carbon Waste Solved by Software Technique
Huge under-utilization of virtualized servers is caused by inefficiency in migrating VMs across servers, to accommodate load spikes. As a result, an immense amount of wasted power and resultant carbon footprint and other ecological side-effects are created. A new software technique overcomes this
Boosting server utilization from 60 to 90% by accelerated VM migration
There is a lot of untapped server utilization resulting from the inability to live-migrate VMs quickly. A technology I've been developing (MemoryMotion™) greatly accelerates migration time and lets server utilization rates be pushed up from 60 to 90%. This enables far greater ROI and improvements in power consumption.
Portable Linux future using LLVM
Imagine a Linux distribution with binaries that adapt to whatever processor you run them on (no 586/686/etc specific versions). That's one of the things that LLVM could enable. Compile once, run anywhere.
Red Hat and others may be good M&A targets after Cisco server market entry
Cisco's Unified Computing announcement makes it clear that data center players must offer converged solutions. A likely quick response will be M&A and deep partnerships. Red Hat is key in the virtualization space.
The next generation netbook concept
To get the next billion people online, netbooks must continue to become less expensive and thus more accessible. An interesting way to achieve this is to separate out the display and smarts into two pluggable units, one which can be shared by many, or replaced functionally with a smartphone, PC, or server.
Future of Linux desktop: co-Linux on Android
We're at the native Linux desktop, moving towards the Android desktop (netbooks coming soon). What would bridge those two environments, is to offer a second Linux lightweight sandbox which runs along with Android. This would give native Linux application compatibility to Android, offering the best of both worlds.
Trusted hypervisors to enable commercial HPC@home services
Volunteer projects like SETI@home successfully crowdsourced scientific computing to a cloud of idle home computers. Trusted bare-metal desktop hypervisors enable scaling this to a commercially viable business model.
Linux kernel needs more modularity for bare-metal hypervisor viability
As the hypervisor becomes the new OS, if the Linux kernel is to compete, it needs a new push to modularize everything humanly possible to provide a minimal footprint to be accepted as "bare-metal".
The future of PCs: pay-to-enable-features?
With hyper-commoditization of PC platforms, the only way forward may be a change in business model where platforms come with extra hardware features which are enabled on-demand for a fee.
The Linux desktop could be profitable soon
While some mainline Linux vendors have left the Linux desktop for dead, this article explains how I believe a new set of market conditions opens the door for profitability.