ISP traffic shaping - cont.

Thought it might be a good idea to update you regarding latest developments and research results since the previous post.

I've been trying the 1.9 beta of uTorrent (available for Windows, runs perfectly under Wine on Linux), since I heard it supports uTP. The reason uTP is important, is because it is believed that ISPs aren't "shaping" this sort of traffic. The experiment seems to be progressing well, as download speeds increased, but still not perfect - not everyone uses uTP-enabled torrent client. But still, now my computer downloads from other computers accross the world, at high speeds, using the new protocol.

So there are few problems which still keeps me from getting my full bendwidth:

1. No open-source implementation of the protocol - slower adoptation.
2. Few closed clients are implementing the protocol, and are still in beta - slower adoptation.
3. ISP still "shapes" other protocols. Annoying.
4. Future problem: will the ISP "shape" uTP?

Happy passover.

Conficker coming to Linux

I guess that by now you've all read about the latest threat over Windows-running PCs - the Conficker worm. So far Conficker had shown some very impressive capabilities, both in infecting innocent machines, and in evading AV products and other security measures.

Today I found at work a Conficker mutation which affects Linux-running computers. Currently it seems to be targeting only specific, debian-based, distros, such as Ubuntu.

Users of older versions of Ubuntu (7.10 and below) are protected due to usage of older kernel.

Everyone who uses newer versions, hurry up and downgrade, or install a more secure and solid OS such as Windows Millenium Edition.

ISP traffic shaping

As I've written before, I'm using file-sharing every now and than, to download a new version of my Linux distro (and other distros for testing), or to download a TV show which doesn't show in Israel, or I cannot watch by other means. Also, as you may all know, the best way to do so, is using a torrent client, such as Transmission, uTorrent, Vuze, etc.

It seems ISPs in Israel (and probably world wide) are playing a double-game. Their first interest is to sell us as much bandwidth as possible. Only recently a new commercial went on-air telling us how much more music and movies can be downloaded if one would double their bandwidth. On the other hand - it seems some ISPs are doing traffic shaping.

Me and a friend of mine are both subscribers of a large ISP, which has the word "gold" in its name, and we are quite sick and tired of this. After doing some thorough tests, we came to a conclusion that torrent traffic is being limited during most of the day. In those hours, the torrent would download at full throttle only from clients subscribed to the same (our) ISP. As for the rest - download rate was roughly 1KBPS. At about 2AM, the limitation is removed, and all of a sudden those same clients started sending data at blazingly fast rates, which lasted until morning.

What bothers me more is the fact there's no transparency regarding traffic shaping. If I'm being limited - I want to know about it. Some uses are legitimate, and consumers must be aware for those limitations. Moreover, what if other services are "shaped" as well, such as the bandwidth-consuming youtube? I like watching youtube videos at HD, and don't want those to be choppy (and I noticed they sometimes do).

So what do you think? Is your ISP shaping your traffic? Also - what can be done about it?

IE8 optional in Windows 7

Is this true? If so, it doesn't mention how could one install a different browser (so you have to use IE8 to download another browser). Interesting.

Either way - this is a good step toward making Windows a safer OS.

What matters in a web browser

I've been installing variety of browsers ever since there were browsers to choose from. I like playing with them, comparing them, and just have them lying around.

Today I've installed Safari 4 beta. What can I say, the Apple dudes delivered what they had promised - a blazingly fast browser, slick UI, and it has the coolness atmosphere that surrounds Apple products. Actually, it feels like the fastest browser I've ever used. Moreover, Hebrew support is great, and many problematic Israeli websites are loaded as perfect as they load on MS's browser - only way faster.
So I played with the cover-flow interface, browsed some heavy websites - and closed it. I don't think I'll get back to it anytime soon (unless for finding some vulnerabilities within). Why?
It seems that since its 0.9 version, I always got back to Firefox after trying different browsers. Chrome was the only browser that stole me for an entire week. Why?

I know the answer. Although it is not the prettiest, not the fastest, nor the best supported browser - it has a HUGE community. And what does this community bring? Free (as in freedom and as in beer) add-ons. Without my ad-block, foxmarks, Firebug, proxy-switcher, screengrab, themes, etc. I'm totally lost. Those add-ons really make the difference! The Internet looks and behaves differently (in a bad manner) without them. So no matter what the other company will bring out the the factory - that browser doesn't stand a chance until it starts supporting FF's add-ons. This reminds a little the hold MS has on the desktop applications - one is so familiar and comfortable with them, that making a switch could be a hard decision.

So what add-ons do you have on your fox?

Lunix is Soviet

This one is simply hilarious. As much as it's old, it's still funny. Not sure whether the writer was joking or is a complete idiot.
And what king of a parent will you be?

Giving a hint

Every major DB I know supports some sort of "hinting" mechanism. What is it and why is it needed?

When a DB compiles a SQL query in order to execute it, it also decides about the best way to execute that query: which data to fetch first, which indexes are to be used, how to perform joins, how many rows to retrieve, etc. The quality of these decisions has major impact over the performance of the query.
As part of this process, a component usually known as the optimizer is running. This component is required for the construction of the execution plan, and sometimes it takes "drastic" measures, such as rewriting the query so that WHERE conditions would be executed in a different order.
Over the years RDBMS providers had put a lot of effort in order to achieve the best possible optimizer. In the course of time it started using statistics (based on ranking mechanisms and previous queries). But before that happened, each RDBMS provided an extension to the query syntax (SQL) so the developer could provide some hint to the optimizer about how he thinks the query should be executed. As the years went by, providers started recommending to avoid those hints, as the optimizer usually did a better job.

I hadn't seen a good optimizer hint for some years now, until this week. I was trying to improve some query in MS SQL Server which lasted forever, and involved an external data source. After I was about to give up and rewrite the code, I decided to give it one last chance. Digging in the execution plans I realized MS's optimizer did awful job in executing the query, so I check their hint syntax. My addition to the code was "OPTION (hash join)", and viola, the query completed within few seconds. Impressive improvement.

Improving the query (by better building it) will always be better than adding a hint, but when all else fails, this might be the only solution.