Sunday, January 29, 2006

Update on IONE NAS

OMG, I'm such a dimwit! Having typed in three paragraphs of text I just managed to press "Discard" button and have to do it all over again! (Or is it Picasa that has censored my post because of coincidental proximity of words "Picasa" and "sucks"?)

Few people have asked whether this device supports NTFS. The answer is, I don't think so. Embedded version of Linux that runs this device (I think?) is very simple. I would very much doubt that it is capable of working with a drive that you have formatted yourself and just fitted in...

Since there's a chance, I'd post a summary of my experiences with the device so far. There are issues :(

It works flawlessly as a backup device, and it is well suited for storing media that can be streamed, meaning, after a file is located, reading from it is performed at a constant, moderate rate. Home videos work fine, and storing your music collection there does too. There is one scenario where performance may be intolerable, it is when you store your digital photo collection on it. Likely due to some issue with Windows drivers for network adapter, or due to some networking configuration in Windows that I was unable to resolve, performance is terrible in both Picasa and Fuji FinePixViewer. Mind you, slideshow would work OK if you set delay between pictures to 5-10 seconds. But to have to wait the same 10 seconds for a randomly selected thumbnail you double-clicked on to open, this is just maddening. I agree it is a typical YMMV case, because my other machine with either Solaris or Linux do not have the same issue, the lag is much more tolerable 2-3 seconds. There's some misunderstandanding between this NAS and my Win XP Home notebook as far as networking is concerned...

Saturday, January 28, 2006

Some resolutions

Do NOT
- try to find best deal on storage until fall
- consider pulling 2x512 sticks of memory out of my Java Workstation and replacing those with 2x1024

Do DO
- continue looking for resources on self-taught patent application filing
- figure out what the heck "Asynchronous Request Processing" is and how to make it work in Glassfish/Grizzly

Tuesday, January 24, 2006

People who type too much

It is not in human's nature to admit one's own flaws. Same with me. Being a slow typer I came to actually appreciate the fact that I have more time to think on a programming problem, or an email. And that I have to express my ideas in a terse form, is a virtue in my eyes. In fact, just recently I was working on changing a bit of twisted and complicated code.

This code deals with localization. Like every other piece of functionality that can be characterized by some likeness of its features, it's a "service". Deployed on its own cluster, and stuff... Like almost everything at this place I'm at. So, this bit is supposed to do the following: accept a Collection of keys, and produce a Map of these keys to Collections of messages. Each message is a tuple, it has some index and the message itself. This way, for any given object there are several localized messages of particular kind. Should I have been amazed? Probably not, I've worked for over a year here. This service is NEVER asked to give translation for more that one object at a time! Now, please realize that all these calls are remote and try figuring how much serialization overhead this over-engineered solution imposes. But forget that, serialization is performed by machines, after all. While I appreciate when I'm being surprised, to spend so much of my time getting down to the core of the matter is sad.

I think constructions like these are direct result of some programmers being excellent typists. Thank chance we are not being paid by the line of code, how much worse would it be then?

Wednesday, January 11, 2006

Epidemiologists' error

I'm not entirely sure of this, but at least one collegue has told me that the flu shot I got missed the strain that's widespread this year... Either that or I'm having a radically nasty cold. Had to stay home one day, hope the worst is behind me. Couldn't do much beyond replying to emails.

But I managed to revive my old machine (Slot-A Athlon 700) to make it a load client. With two notebooks and with this one, should suffice to load Grizzly properly (I was advised that I need ~2000 concurrent connections to see it shine). Also I'm considering squeezing more concurrent threads from each JMeter instance, by reducing the stack size. Will see if that makes any difference...

Monday, January 02, 2006

Intel agony

I may be drunk but I'm not out of my mind. Two Photos from Intel Corporation's Construction of Their Booth. How does one qualify tils?! What is the lowest point?! Should we salivate and ...bate over a bunch of pixels? Wow. Wow. I'm speachless. Way to go!

Disclaimer: writer of this is a (modest, really, 500 shares since Nov. 2004) investor in AMD. And I intend to hold on to it for another year.

Sunday, January 01, 2006

Death by a cookie

The sky is falling! Not for everybody, but for a big segment of Internet business, for sure. As the story reported here among other places develops, people will come to realize how much their privacy is invaded by those tiny one-pixel tracking images used by web analytics service providers to collect information about users' behavior on the Internet.

What is the problem here? Well, it appears that as more and more websites get wired to collect data about web advertising performance, or to (innocently enough on the surface) analyse web site usability, either the web analytics service provider, or a third party capable of intercepting unencrypted traffic between Internet user and a web site using such tracking technology, is capable of recreating user's full experience. I find it obvious and unneccessary to elaborate why tracking someone's activity on the web is unwanted.

Who do we blame for this? At the root, it's reckless disregard for Internet user's privacy, multiplied by greed on part of web analytics service providers, and by lazyness/incompetence of their customers (usually those are marketing departments wooed by sexy reporting such service providers can offer).

Why is there blame to assign? Primarily, because source data and tools for reporting on that data are readily available to IT departments of web site operators. Analysing e.g. Apache access logs would not cross the boundary of different web domains and will guarantee users are not tracked globally.

Who's involved? Besides the already mentioned web anlytics service providers (WebTrends, <any>ture), and obvious pre-GOOG web ad resellers (<whatever>media?, <N-th>click?), the "inherently unevil" GOOG is in the number. How come? simply, any cookie planted at some point to be sent back to domain google.com will track us poor souls all the way across Internet. Asking questions of google while doing your research on an invention you're were about to patent while being logged on to google account? Reading news from websites of dissenting talk referred to by Google news? Planning trips using google local? Bad luck... It all can be accounted back to you, easily.

Can users do anything to fight such horrifying privacy invasion potential these technologies bring? Sure... kind of.

There's an option of simply disabling third-party content on loaded web pages (one would have to disable all kinds of plugins, too, as plugins are not bound by browser settings in this regard). Negatives of this approach are overwhelming: some web sites are not useable at all with plugins disabled (really, those sorry asses making their websites done as Flash movies and with no option to get to content in any other way had to be punished someday... is it now?) Besides, web sites that use egde caching technology (usually delivered by Akamai), are not useable after such a change, since images (and other static assets) are referred to by URLs that do not belong to domain of the web page itself.

There are various solutions on the market claiming to address these concerns. Issue here is basically one of trust. It can be quite hard for a user to distinguish spyware package in disguise from an honest privacy protection solution. Besides, the way web sites are made these days, as I explained just above, usability will suffer anyway.

Disclaimers: the above represents my personal (as much educated as it can be) opinion and is not guaranteed to be 100% correct, while I believe it is so. I hold no interest (including short) in any of the companies mentioned.

UPDATE: New web analytics implementation scandal involving Apple's iTunes, all over the Web. I wonder if the iTunes mini-store shares cookies with IE on Windows and Safari on Mac OS... Could test this later if find nothing better to do.

Funny thing, this is not new at all. In a book by Mr. Siebel I'm reading right now, he recalls one of Web pioneers as saying that cookies will be the death of the Web. I think I understand what he meant. And I would disagree with those Sun Microsystems founders who say "There is no privacy, get over it". I believe there are technical means of achieving the same goals without sacrificing privacy, at least as far as browsing from the privacy of one's home is concerned.