Antimail : Click or miss

Statcounter: Bing leapfrogs Yahoo again

AdiOltean — Sat, 27 Jun 2009 11:00:24 GMT

You might remember Statcounter.com as the site that came in the news a few weeks ago when Bing overtook Yahoo, due to the sudden interest immediately after launch. Back then, as many predicted, it didn’t last long, as people quickly switched back to their old search habits.

But I just checked the Statcounter site and noticed that Bing just got ahead Yahoo again. Granted, this time in North America region only. On last Friday (26 Jun) Bing had 8.74% query share, compared with Yahoo with 8.55%.

Of course, this situation probably won’t last long either, but there are a few observations to make:

1) The two graphs are getting closer and closer, and it’s probably safe to predict that they will start intersecting a lot more often in the next weeks. The same trend can be seen in the WW numbers as well.

2) There is a curious pattern when Bing traffic is higher on Wednesday and Friday (not sure why?)

Either way, it’s an interesting area to watch out in the next weeks to come …

Bing Community - an undiscovered site

AdiOltean — Thu, 25 Jun 2009 08:29:00 GMT

Check it out:

http://www.bing.com/community/

Nice layout & content organization. Much better organized when comparing it with the typical forums/community sites that you see on Microsoft sites or MSDN (or other non-Microsoft ones)

FCI – how to use classification (video)

AdiOltean — Thu, 28 May 2009 23:23:31 GMT

Just found a new video demonstrating the new classification feature in Windows Server 2008 R2. Enjoy!

Special thanks to dawho1 who posted the video.

Mommy, Why is There a Server in the House?

AdiOltean — Wed, 09 Jan 2008 09:58:39 GMT

I love this - if you have kids, and if you are a geek, you'll understand.

http://gizmodo.com/photogallery/microserveces08/1000446145

Puzzle solution: Xen voting algorithm

AdiOltean — Tue, 27 Nov 2007 00:19:38 GMT

I think that the problem stated in one of my earlier posts is one of the most fascinating puzzles I came across recently. Many people that got confronted with it said bluntly that the problem simply has no solution, otherwise it would contradict common sense, information theory, etc. But surprisingly, it does have a solution.

For X% strictly bigger than 50% the solution is known as the MJRTY algorithm, discovered a while back by R.S. Boyer and J.S. Moore (also called the Majority Vote algorithm). Although the algorithm is deceptively simple, its proof is not. the In the PDF cited above, the authors present a complete proof and also some interesting history around this non-trivial result.

The correct algorithm was also found by Lailin Chen - also presented in the post comments, but without a proof:

Well, Since we already knew there is one and only one who has more votes than any others, we can find him by letting the votes "fight" each other:
Picked the first one, say X1, count 1 for X1, then move on, if X1 repeats, add count to X1.
When X1 doesn't repeat, reduce the count by one, if it reaches 0, picked up the current one (say X2), and do the some counting to X2.
Keep doing this until the end. The last one left with a counting bigger than 0 is the leader:
X3 X1 X1 X2 X1 X3 X3 X3 X4
   ^
X3, 1
X3 X1 X1 X2 X1 X3 X3 X3 X4
         ^
       X3, 0
X3 X1 X1 X2 X1 X3 X3 X3 X4
               ^
            X1, 1
X3 X1 X1 X2 X1 X3 X3 X3 X4
                     ^
                   X1, 0
X3 X1 X1 X2 X1 X3 X3 X3 X4
                           ^
                          X1, 1
X3 X1 X1 X2 X1 X3 X3 X3 X4
                                 ^
                               X1, 0
X3 X1 X1 X2 X1 X3 X3 X3 X4
                                       ^
                                    X3, 1
X3 X1 X1 X2 X1 X3 X3 X3 X4
                                             ^
                                           X3, 2
X3 X1 X1 X2 X1 X3 X3 X3 X4
                                                    ^
                                                  X3, 1
So X3 is the winner.

Here is a more formal description of the algorithm:

- You have with two registers: an “candidate register” containing the name of some candidate (candidate), and a counter (counter).
- Candidate_register = NULL;
- Counter = 0
- For each new candidate:
     If (Candidate_register is NULL)
            Candidate_register = candidate
            Counter = 1
     Else If (candidate == candidate_register)
            Counter++;
     Else if (counter > 0)
            Counter --;
            If (counter == 0)
                Candidate_register == NULL
- At the end you have the correct candidate in the register.

For X < 50% we can adapt a more complex variation of the same algorithm which requires two passes instead of a single one. Instead of keeping a single register (keeping the last candidate) and its associated counter, we keep a number of K registers (with their associated counters) and run the same algorithm. The integer K is chosen such that 100/(K+1) < X. After this pass, we discovered K candidates but of course the candidate with the biggest counter might not be the real candidate. To find the top candidate we simply reset all counters to zero (but keep the candidates found) and re-run the same algorithm.

- Instead of maintaining one “element” register and one counter as in the main algorithm, you need to use K registers, each with K counters.
- First pass
     Initialize all candidate registers with NULL and counters with zero
     For each candidate in the log file
          If the candidate is in one of the K registers, increase its counter
          Else if there is one candidate register which is NULL, put it there and increase its counter with 1
          Else, reduce all counters by 1
               If some counters become zero, set their corresponding candidate registers to NULL
     At the end, each of the candidate registers contains every element with frequency > 100/(K+1). One of them is the winner
- Second pass
     Initialize all counters with zero, but keep the candidate registers filled with the values above.
     For each candidate in the log file
          If the candidate is in one of the K registers, increase its counter
     At the end, you have all the correct counters for all the candidate registers.
     Select the one that has the maximum count.

A variation of this second algorithm is briefly mentioned in this presentation although I could not find a formal article describing it.

What all these algorithms have in common? The fact that you know ahead of time that there is a winner, but you don't know which one. While the algorithms are simple to implement, this is a pretty strong condition that is not likely to be encountered in our real world. Maybe on a planet called Xen :-)

Zune 80 - $400 on eBay?

AdiOltean — Sat, 17 Nov 2007 02:44:44 GMT

A few completed transactions:

- $420: http://cgi.ebay.com/Microsoft-ZUNE-2-Gen-Black-80-GB-mp3-NEW-NIB-EXTRAS_W0QQitemZ320182867203QQihZ011QQcategoryZ147175QQssPageNameZWDVWQQrdZ1QQcmdZViewItem

- $405: http://cgi.ebay.com/Microsoft-ZUNE-2-Gen-Black-80-GB-mp3-NEW-NIB-EXTRAS_W0QQitemZ320183303981QQihZ011QQcategoryZ147175QQssPageNameZWDVWQQrdZ1QQcmdZViewItem

- $399: http://cgi.ebay.com/Microsoft-Zune-2-Gen-Black-80-GB-20000-Songs-Dig_W0QQitemZ320182491104QQihZ011QQcategoryZ147175QQssPageNameZWDVWQQrdZ1QQcmdZViewItem

Is my harddisk (almost) dead?

AdiOltean — Tue, 13 Nov 2007 21:03:30 GMT

I just started to play with HD Tune, just to get a tool to look at my S.M.A.R.T. data. Not that I trust SMART a lot, but I wanted to see what's there.

So that's how I discovered my main drive has a bunch of reallocated sectors. Ouch!

What is a reallocated sector? Whenever a sector becomes "bad" for some reason, modern harddisks are "remapping" this sector to some other location of the disk. That's why you almost never see "bad clusters" on a modern harddisk. See for example the result of the CHKDSK command on the same drive - it says that everything is fine:

C:\Windows\system32>chkdsk c:
The type of the file system is NTFS.

WARNING! F parameter not specified.
Running CHKDSK in read-only mode.

CHKDSK is verifying files (stage 1 of 3)...
118464 file records processed.
File verification completed.
129 large file records processed.
0 bad file records processed.
2 EA records processed.
75 reparse records processed.
CHKDSK is verifying indexes (stage 2 of 3)...
460369 index entries processed.
Index verification completed.
5 unindexed files processed.
CHKDSK is verifying security descriptors (stage 3 of 3)...
118464 security descriptors processed.
Security descriptor verification completed.
16840 data files processed.
CHKDSK is verifying Usn Journal...
35955328 USN bytes processed.
Usn Journal verification completed.
Windows has checked the file system and found no problems.

244196351 KB total disk space.
104969600 KB in 101394 files.
     50680 KB in 16841 indexes.
         0 KB in bad sectors.
    227651 KB in use by the system.
     65536 KB occupied by the log file.
138948420 KB available on disk.

      4096 bytes in each allocation unit.
61049087 total allocation units on disk.
34737105 allocation units available on disk.

There are two problems with reallocated sectors. First, each harddisk comes with a limited "pool" of empty sectors that can be used as reallocated sectors. When you run out of those, then the automatic protection goes away, so you will start seeing more and more bad sectors at the OS level.

Second (and most importantly) there is a performance problem with reallocated sectors. Due to the fact that some sectors are remapped to another area on the disk, sequential I/O on those sectors is getting randomized (becomes random I/O) with very different performance characteristics. How big is the performance impact? Pretty big. Let's say that you have a no reallocated sectors in a 40 MB interval. If you want to read 1 MB in this interval, you will read it at a standard sequential I/O rate, say about 50 MB/s on a regular SATA disk. So reading will take 1/50 = 20 ms.

Now, let's pretend we have a reallocated sector in the middle of the 1 MB that we want to read. The reading time will include those 20 ms above plus the two additional seeks. Since a seek is usually around 8-9 ms (or even more) for a standard SATA harddisk, you get around 40 ms for reading the 1 MB which is twice as long. So, from a bandwidth perspective, you are reading that 1 MB at 25 MB per second, which half of the actual speed. So what might see is a sudden decrease in sequential I/O bandwidth whenever there is a reallocated sector around.

The interesting thing is that you can "spot" the approximate location of reallocated sectors by doing a sequence of sequential reads over the entire harddisk (from the beggining to end) and see where you have sudden drops in I/O performance. Fortunately, the same tool mentioned above - HD Tune - has a benchmark mode which allows precisely this.

Here is the picture of my backup drive (a Hitachi 250 GB drive):

You can see that the transfer rate decreases smoothly from 60 MB/s (near to the outer region of the disk) to 30 MB/s (near to the center of the disk). There are no sudden drops. In contrast, here is a disk with a lot of reallocated sectors:

You can see a lot of drops along the blue sequential read path, with a variability of about 50% in some cases.

Note - if you are doing these sequential I/O tests, make sure that your harddisk is not in use by any other applications or the page file - this will cause additional seeks which would "pollute" the original graph.

For that reason, I haven't run yet the benchmark on my main drive since I know that paging will distort the results - I'll try this later ...

[Cool spam] Our internal discussion list is nominated in a Who's Who award!

AdiOltean — Sat, 06 Oct 2007 00:39:25 GMT

Here is an interesting piece of spam I received this morning on our internal discussion list:

Dear VSS Lab Account,

    You were recently appointed as a biographical candidate to represent your industry in the Madison Who's Who Among Executives and Professionals, and for inclusion into the upcoming 2007-2008 "Honors Edition" of the registry.

     We are pleased to inform you that on October 4th, your candidacy was approved. Your confirmation for inclusion will be effective within five business days, pending our receipt of the enclosed application.

     The Office of the Managing Director appoints individuals based on a candidate's current position, and usually with information obtained from researched executive and professional listings. The director thinks you may make an interesting biographical subject, as individual achievement is what Madison Who's Who is all about. Upon final confirmation you will be listed among thousands of accomplished individuals in the Madison Who's Who Registry. There is no cost to be included.

     We do require additional information to complete the selection process and kindly ask that you access this form on our website:

Click here to register

Or you can manually enter this address into your web browser:

http://www.madisonwhoswho.net/basiclisting

To ensure your biographical data is received in time, please complete the online form above as soon as possible. Our editorial deadline is quickly approaching.

Sincerely,

Matthew Johnson
Managing Editor

Madison Who's Who is not associated or affiliated with Marquis Who's Who or any other Who's Who.

Madison Who's Who, Inc. 30-01 Northern Blvd. Long Island City , NY 11101 USA

Great - so this would be our first occasion to promote our internal, super-secret VSS-related discussion list in a public Who's Who entry.

After I checked that the links are indeed not pointing to some sort of phishing-attack server, I went there and filled the forms - maybe we'll get approved!

I'll keep you in touch with the progress...

[RO] Seminar - Storage si programare distribuita

AdiOltean — Thu, 03 May 2007 20:57:00 GMT

Florin Lazar, Todi Pruteanu si cu mine am demarat in cursul acestei saptamani un seminar tehnic pe teme de storage in Vista si programare distribuita in WCF. Mai multe detalii aici: http://reg.studentclub.ro.

Pana acum seminariile din Bucuresti si Cluj au avut un raspuns fantastic! Daca aveti opinii, impresii (pozitive sau negative) m-as bucura sa le puneti in sectiunea de comentarii...

Si nu in ultimul rand, multumiri lui Todi si Microsoft Romania pentru organizarea acestui eveniment.

Windows Home Server - links

AdiOltean — Fri, 16 Mar 2007 01:43:27 GMT

A friend of mine didn't know about Windows Home Server - so I sent him these links:

- Main site: http://stopdigitalamnesia.com

- Interview with Charlie Kindel http://channel9.msdn.com/ShowPost.aspx?PostID=270965

- Download: https://connect.microsoft.com/site/sitehome.aspx?wa=wsignin1.0&siteid=38

- Articol on Wikipedia: http://en.wikipedia.org/wiki/Windows_Home_Server

- Blog: http://blogs.technet.com/homeserver

- Press release: http://www.microsoft.com/Presspass/press/2007/jan07/01-08WindowsHomeServerPR.mspx

Demo: New virtualization technologies in Windows Server (Longhorn)

AdiOltean — Fri, 02 Mar 2007 23:32:16 GMT

A nice demo that shows you several things:

- Windows Server virtualization (a separate partition running a striped-down version of Windows to manage all other virtual machines)

- Running an 8-proc virtual machine

- 64-bit and 32-bit machines running concurrently

- Managing virtual machines with System Center (nice UI)

- System Center Operations Manager: monitoring the VMs, provisioning new hardware in a scriptable manner.

Video: Longhorn - Windows Server Virtualization

Previous Versions in Vista - on Channel 9!

AdiOltean — Tue, 27 Feb 2007 05:29:48 GMT

Hot and fresh from Channel 9: a session on Shadow Copies in Vista, and its applications: Previous Versions and the brand-new System Restore.

Check it out.

The intentional slowdown

AdiOltean — Fri, 09 Feb 2007 11:57:09 GMT

The Daily WTF has a funny story around intentional slowdowns. I have one story to share as well.

Back in the college days, when we were still busy with exams and courses, some of my friends started working at various companies in our city (Bucharest).

One of them managed to get a contractor position at a certain software company. His job was to take a certain application developed by that team (let's call it "Standard"), and to create an "Advanced" version of that application, since some customers were willing to pay to upgrade to a more powerful version of the same app.

After a while, the "Advanced" version of this application was presented to customers and they were delighted. Indeed the application seemed to be more powerful than its "Standard" counterpart - it certainly spent more time "thinking" when it performed certain tasks.

What was the secret behind the Advanced application? After a couple of beers, my friend told me: a bunch of inserted Sleep() calls.

P.S. I haven't heard from my friend in a while - his contact with the software development industry was very brief.

Flicks - A gesture based scrolling interface in Vista

AdiOltean — Wed, 31 Jan 2007 11:45:11 GMT

If you have a Tablet PC or a touch-sensitive model, check out Pen Flicks. Here is a video capture of Pen Flicks in action:

I just found this nifty little feature after installing Vista on my old Tablet PC and it works great. It works on any list box in any application (Internet Explorer, Outlook, Shell, etc)

P.S. I heard that the Apple iPhone has a "revolutionary" scrolling interface based on flicks - hmmm... what was their source of inspiration?

The evolution of customer support

AdiOltean — Sat, 09 Dec 2006 11:12:00 GMT

- 2000 - You call their customer support. “Wait a second – did you actually install an application after installing our sound card? We don’t support this configuration. <hang>” (and that guy will likely get promoted for solving most support calls in record time).

- 2001 – You call their customer support. Nobody answers because the company is out of business due to the .com fallout.

- 2002 – You call their customer support. But you get someone with a think non-English accent which keeps asking you about completely unrelated stuff.

- 2003 – You finally figure out that if you put your problem on the newsgroups you will get help.

- 2004 - You complain about the problem in your blog, and you eventually get suggestions in comments section.

- 2005 - You complain about the problem in your blog, get instant notoriety, and wait for someone from that company to reply in the comments section :-)

- 2006 – You post a video on YouTube with your problem, get instant notoriety, and wait for someone from that company to reply in the comments section.

An example in the last category: Here is a guy describing his problem with the Xbox 360 wireless headset.

http://www.joystiq.com/2006/12/09/todays-hottest-game-video-360-headset-woes/

Today's most-watched YouTube game video shows a gamer who is unable to connect his original Xbox 360 controller and Wireless Headset together on channel one. The demonstration implies that this is a widespread issue with no fix yet from Microsoft, although we haven't had any issues.

The gamer shows the process of activating the headset and having a conflict with his original controller. He then repeats the steps with a new, extra controller and has no problem.

It's like the I-Team without having to sit through the rest of the made-up local news. See the video after the break.

I don't know the solution to his problem, but I notified the XBox team anyway... I hope that they will respond!

[Submit this story to digg.com]

P.S. By the way, for any problems, please call the Microsoft support - they provide an excellent service, (as they always did), won't hang on you :-), and if it is our problem, the call is free.]