Author Archives: Paul

TP-Link MR3040, autonomy test

Posted on by
3

As promise, I’m looking for MR-3040 information about autonomy. This device is really interesting as it provides an internal battery to power the device on. The question was, how many time can we use it, is the autonomy good enough to have a real mobile device you don’t have to care about autonomy when bring out to public !

The first thing is to reload the battery, from 0 to 100% it takes 3:10 using the given equipment.  The next thing is to empty the battery by just doing nothing else than switch it on, as a result, the MR-3040 ran during 7hours. This is a really good result as it means that we can bring it out most of one day (or night).

The last step is to empty the battery transferring file from & to to mr3040, in this scenario, the battery autonomy should be smaller but the result is an autonomy of 5 full hours with a transfer rate about 1.5MB/s. I was expecting a shorter time so it is a good news.

The main issue I can raised after this test is the coverage area of these kind of devices (MR3040 or MR3020) really limited to a some meters and really blocked by rock walls. This is limiting the share with the neighborhood capability. It makes it more recommended to have a share into a room.

Installation of a piratebox on T-Link MR3040

Posted on by
14

I just received the TL-MR3040 wifi portable access by T-link. This little box is really like the mr3020 i had for my piratebox with the main advantage of having and internal battery making the piratebox mobile for a small price ! For actually 38€ (but you may found it at lower price) it is an interesting and compact solution to bring you piratebox out.

I’ll try in the next part of this article to detail the installation process as friendly as possible. Never forget that this kind of operations are not always easy and may damage your device. Do it at your own risk.

Continue reading →

Nmon cpu_summ vs cpu_all in a SMT environment

Posted on by
Reply

In a SMT CPU you have real primary core and SMT core, all can have process allocated to them but the performance of the primary logical processor compared to the SMT one are not equivalent. Usually the SMT power is about 20-30% of the primary one.

This impact the way the performance is measured

Continue reading →

Network performance measurement script

Posted on by
Reply

I regularly have to test some network products like routers or homeplug devices and most of the time, I like to test the performance of that product. I’m doing it generally a couple of time and the result can vary test after test and I did not had a good, simple and basic tool to use to measure this variation.

So, after having think to it for long months, I finally started to write some bash lines to get this tool and start to get measures…

Continue reading →

Oracle DB performance over low latency networks

Posted on by
Reply

Here is an interesting article on the impact of the latency on some usual Oracle SQL statement over Local, LAN and WAN networks.

http://markbairden.blogspot.fr/2012/03/database-performance-measuring-effects.html

Wan latency uses is 12ms which is small for this type of connection. Impact is huge with a factor of 10. To be taken into account in your designs !

Thank you Mark for this article !

New low consumption Atom devices

Posted on by
Reply

Not really a news, but an interesting update on two new devices sold since march 4th, Atom family have, at least, two new Core, the first one is the N2650, the second one the N2850, they are two dual core, 4 thread CPU for netbook with a really low TDP.

The N2850 offers a 2GHz system with a 10W TDP, the N2650 offers a 1.7GHz system with a 3.6W TDP.

This has to be compared with my current reference (making no sens for you … but …) the D525 offering 1.8GHz for 13W. More over the IGP has been push from 400MHz to 640MHz on the N2850.

This configuration sound really interesting to build small and fanless machine… now we need to see the first box including these chips.

See Atom family on wikipedia : http://fr.wikipedia.org/wiki/Intel_Atom

Why behind concept PirateBox is just an experiment ?

Posted on by
1

After a couple of days hacking PirateBox based on MR3020 i’m really happy to give this concept life, but behind the announcements like “share freely for less than 40€ with your neighborhood” the reality differs.

The first main thing is that 40€ is the price, port excluded to get the MR3020 router, then you must add a storage (25€ for 16GB), add a battery if you want to be mobile (50-100€). Which made a big difference as it becomes quite as expensive as a tablet PC where you can install the software and be mobile with higher storage, higher mobility and much more capabilities.  So, no magic around this.

Ok, i’ll tell me you can reuse an old external hard drive and not being mobile, just switching it up at home… and you’re right, this is a good capability for a reasonable price. But, once again, you’ll have to deal with the low Wifi power offer by this low cost router. I mean you’ll be able to exchange with your direct neighborhood : the one you cross daily in the stairs. Why not, it sounds good …

That last argument I also would like to share to finish to be a party killer, is about the Hadopi stuff and piracy freedom feeling…. Are you mad guys ? As much as I know, getting address of pirate is not so easy when you are downloading on Internet, even if it has been simplified, you’re one on million and the chance to be kept is small, then you will be identified more as a consumer  than a provider. To identify the owner of the PirateBox is is quite easy by triangulation, then you just need to request a judge to get the needed paper to catch a provider..

In my opinion, PirateBox has to be mobile to not being kept or need to not be owned and hosted on public domain and sourced on solar power. (Like the USB Stick embedded in walls, but for more money) Which would be a great and interesting thing but requires investment.

For all these reasons, I love the idea, I like the product in a technological point of view, interesting and easy hack, but I do not consider it as something really ready for mass. It’s hype, it’s geek, not consumers.

Jar file preloading before a go-live

Posted on by
Reply

Sometimes, when you go live an application handling a big applet you could have not network problem as if all the user try to get the associated Jar files in parallel this could saturate your network and impact your go-live.

One of the solution is to pre-load these JAR files into the browser of each user. Unfortunately is sound not possible to copy the files into a specific directory to do it : the Jar files must be loaded for the server.

Continue reading →