I want to know why there is a big difference between PSI scores of mobile site and desktop site
Google Page Speed Insights uses different parameters for desktop vs mobile testing.
For desktop the connection speed used is not throttled, it will load as fast as a web browser will allow (i.e. limited by round trip times, maximum simultaneous downloads per domain and CPU processing power of a decent PC).
For mobile however Google tries to simulate a real world example of a mobile phone on a 4G connection.
4 x CPU slowdown as mobile phones are less powerful and network throttling that includes slow upload and downloads and more importantly simulates a high Round Trip Time for each resource request (higher latency).
Related
I've been trying to understand what a Trusted Execution Environment is and how they work. Why is there such a strong emphasis on mobile devices? I've been trying to look for what's the difference in personal computers versus mobile devices with respect to a TEE. What am I missing?
Even though it’s late; I will add my comments in simplest possible way for reference.
As the world starts to move toward Enterprise Mobility, using mobile devices for work starts to become essential for different companies and organizations. From there a need come up to secure that devices, not only the data, but the processes and memory allocation as well; Especially when governments and sensitive departments start to use mobile devices.
Starting from the very low level of mobile devices architecture, every mobile device has a processor, processor manufacturers come up with new technology which creates two isolated areas running at the same time (e.g. ARM Trustzone) on the CPU and controlled by SoC (Software on Chip).
First area is where everyone use on mobile devices (Normal World / Rich Execution Environment - REE), the second one is the secure area (Secure World / Trusted Execution Environment - TEE). Each area has its own operating system running at the same CPU but their processes and memory allocation are totally separate.
Many mobile device manufactures (e.g Samsung), start to utilize that area, by loading third party secure Operating System (OS) into there (e.g Kinibi OS from Trustonic).
Developing applications (Trusted Application - TA) in the secure world is not easy process, provisioning them there is another story and integrating that applications with the normal world is another story as well (Some sort of especial SDK provided by TEE OS owner has to be used).
It is worth to mentioning that applications running in the TEE can have extraordinary privileges and normally TEE OS Owners limit what TA’s can do.
Lastly, although TEE is considered a secure area for sensitive processes (So far). There are other ways to achieve same level (or even better) of security on mobile devices.
I have a raspberry pi that is running a node server (that receives maybe 10 requests a day) that is running constantly. I am wondering if anyone know about how much bandwidth this will take up or how I could go about calculating the amount of bandwidth it is using.
That's kind of like asking, "How much gas will this car use?". It depends entirely what you do with it and how you use it. Sitting in the garage, it doesn't use any gas. Racing on the autobahn at 200km/hr, it uses a lot.
A server all by itself barely uses any bandwidth. The Raspberry Pi has a deamon that checks with an external ntp server to keep the local clock accurately synced with the outside world, but other than that, it likely uses no bandwidth by itself.
So, any other bandwidth comes entirely from the requests that are sent to it and the responses that you send back. If you measure those and then multiply by how many times per day you make those requests, you will pretty much have the bandwidth consumed. If there are just regular http requests returning a small amount of data, then the total bandwidth is probably less than browsing a few web pages with big images on them or certainly less than watching a few YouTube videos.
You should also be aware that if your server is discoverable on the internet, you may also get some requests from unsolicited sources, either search crawlers or bots looking to do some mischief.
If you want to just measure what your Raspberry Pi is using, there are a wide range of tools for doing that on Linux and you'd have to investigate which ones are already available for the Pi or which you could recompile to work on the Pi's ARM chip.
I want to know what is the difference between iperf tool and Ixchariot tool.
Both are used for performance testing. I want to know difference and features of both tools.
Thanks,
Devanathan.D
Both iperf and IxChariot do network performance monitoring, much in the same way that GIMP and Photoshop both manipulate images. IxChariot is a commercial, enterprise-level, customizable tool. On the other hand, iperf is open source freeware and as such, you get what you pay for. The main advantage of iperf is that it's free. The application is a simple command line executable which can act as either a server or client, and is available on a variety of platforms. There is also a Java-based GUI called jperf which is convenient in that it presents all of iperf's features without requiring the user to recall command line arguments. It seems that jperf may no longer be maintained and some have reported difficulty making it work with more recent versions of iPerf, particularly iPerf3.
IxChariot is more advanced tool that can instantly assess network performance, including wireless performance. Performance endpoints run on mobile, PC, Mac or in any hypervisor or cloud service and allows central management of any platform. It also delivers full application emulation and key performance metrics, including throughput, packet loss, jitter, delay, MOS, and OTT video like Netflix, YouTube.
Some can get by with iperf and similar freeware, just like some can get by with GIMP instead of Photoshop. However, just like most serious graphic artists will eventually migrate to Photoshop, most organizations that really need robust network performance monitoring will eventually turn to commercial, enterprise grade solutions like IxChariot particularly if they need endpoint hardware appliances, advanced UDP, multi-codec voice, video, configurable mixed traffic, configurable TOS/DSCP or mixed traffic with different COS (TOS/DSCP).
Full disclosure - I am an Network Engineer at Ixia.
iperf is free. Thats the main difference to be truthful
I tried running a Meteor.js app on my Samsung Smart TV (2013 I think) stock Internet Browser and received "out of memory errors". I'm interested in see if any can or has run a Meteor.Js app on some of the latest SMART TV Browsers...
eg. I tried a running a sample app to test Meteor.js Streams package...
Just curious...thanks for any feedback!
I am doing some work against LG Smart TVs and have been debugging out-of-memory errors as well. Two major things: large client side minimongo databases are all in-memory and can contribute to the issue. Usually just limiting how much you publish on the server should help. Secondly, images. Those are likely your main culprits. These TVs don't understand single page apps too well and will only flush memory consumed by images when the page reloads.
My company has a Compact Framework.NET WinForms application which runs on rugged handhelds manufactured by companies like Motorola, Intermec and Psion. These are expensive devices with built-in barcode scanners that are used in harsh conditions.
The configuration of the handheld application is managed by business users through our web site. The devices pick up the configuration when they sync from within the handheld application. Field workers use the handhelds, business users use the web site.
The business users have expressed the desire to, for lack of a better description, configure and preview or even fully use the actual handheld application through a web browser. They want to make configuration changes in the web site and immediately see what the impact will be in the handheld, without having to have a physical device (again, the devices are quite expensive). They want to be able to create training materials or conduct sales meetings and be able to demonstrate the application to their customers without having a physical device on hand.
Microsoft offers several Device Emulators, but they are probably too complex for business users. They are developer tools. One idea might be to somehow use the emulators within virtual machines possibly in conjunction with Terminal Services or even some kind of clever screen capture/VNC to show an emulated device in a browser. I suspect running emulators in the fashion may not exactly be a scalable solution, however. Also, only one emulator at a time on a single machine can be "cradled" and connected to network.
I'm looking for any suggestions which might help me meet the business users' requirements.
Thanks.
The only thing I can think of offhand is not that simple, but would probably be useful (and certainly the only "true" way for them to test).
I'd create a service that works like the Remote Display app (part of the WinMo Developer power tools, also ships with Platform Builder for CE), in fact it might just use that app (the source code for it actually ships with Platform Builder, so the eval version of PB would get you that source).
You would then create a web interface that acts as a "shell" for that service, marshalling the display image out to a web page and image clicks back as mouse events to the device.