Comments on David Pogue Article

David Pogue is not a computer nerd. I mean this neither as a compliment nor an insult, just a statement of fact. I mention it because when reviewing computer "things" his mind set is different from that of a computer nerd.

Mr. Pogue plays with toys and writes about it. This is certainly useful, as far as it goes. I read his columns and have learned things from them. But without an IT background, Mr. Pogue is limited in what he can add to the discussion above and beyond what is in front of him. That is, he can't put things into perspective. It's one thing to say what a device or service or software is, but quite another to say when/where/how to use it. (I can describe a scalpel, but you don't want me operating on your appendix).

With this in mind, let me try to put his article in today's New York Times, The Cartridge, Updated, Catches Up to Data, in perspective.

The article reviewed three removable cartridge devices that are fast and hold lots of data. In general he liked the devices but felt the cartridges were too expensive.

What he failed to mention is that each of these removable cartridge drives is a single point of failure. That is, if the drive itself dies, you lose access to all your data.

Experience tells us that computer backup devices have a limited shelf life. It is all but guaranteed that in a year or two none of the devices he wrote about will still be on the market. And, it goes without saying, that hardware breaks. So anyone planning on depending on these things, needs to buy two - one for now and one for the future when they can't be replaced.
In addition, the drives should be stored in different physical locations.

And if you are really going to depend on them, a case can be made that you need to buy three drives. In other words, you need to treat the backup hardware itself just like the data. Backup. Backup. Backup.

Switching focus to the cartridges, two of them have no track record. Mr. Pogue mentions the "click of death" that haunted previous backup devices from Iomega and notes that their current product, the Rev, seems to have stood the test of time. An excellent point and one that should make anyone wary of the two newer products.

If one of the newer products sounds appealing, let me suggest making a couple phone calls. DriveSavers and OnTrack are the companies of last resort when it comes to extracting data off hard disks. Call them to see if they will extract data from a Rev, GoVault or RDX cartridge. If not . . .

Posting viewed: times

WSJ article on running Windows on a Mac

This is a review of an article that appeared in the Wall Street Journal on March 20, 2007 called Apple Opens Doors by Running Windows by Nick Wingfield (only available to wsj.com subscribers). The subtitle is: Ability to handle Microsoft operating system may help macs make some inroads.

---

The article says that you need to buy a copy of Windows to use on a Mac. However, it does not say which version(s) of Windows. There are many versions of Windows as we all know, except apparently the author, who repeatedly refers to "Windows" as if it were a single thing without different versions/editions.

To fill in the blank, the Parallels virtual machine lets you run pretty much all versions of Windows on the Mac, even Vista. In fact, it also supports many versions of Linux and even other Operating Systems such as OS/2. The Boot Camp feature from Apple is very different. It only supports Windows XP Home and Professional. No Media Center edition. No Vista. In fact, it only supports the SP2 versions of XP. And, it only supports the more expensive "full" version of Windows XP, not the cheaper "upgrade version". As of March 23, 2007 CompUSA was selling the upgrade version of XP Home for $100 and the full version for $200. The upgrade version of XP Pro was $200, the full version was $300.

The article refers to VMware as new "virtualization" software and refers to Virtual PC as older "emulation" software. They are, in fact, both virtualization software and compete directly. Not to be too nerdy, but they both do emulation.

The article says both Parallels and VMware on Macs take advantage of Intel chips to make it easier to run Windows. A little of this is true, but it gives the wrong impression.

First, from the user point of view, the process of running Windows is the same whether the software takes advantage of Intel chips or not. There are new features in Intel processors specifically designed for virtual machine software. And giving hardware assistance to virtualization software is also done by new processors from AMD. If you have one of these new processors, and a version of virtualization software that is capable of exploiting the new features, then your virtual machines will run faster. What is made "easier" is the programming job of VMware and Parallels as the processor can now do some of the work the software used to have to do.

Omissions

OK, we're running both Windows and the Mac OS on the same computer. Can they share files?

What about the copy of Windows you run on the Mac? Does it have to be a new copy? Can you take the existing copy on an existing computer and run that on the Mac? If this interests you, read about the transporter feature of Parallels.

The terms Host Operating System and Guest Operating System are not used, let alone defined. They are necessary terms when discussion virtualization software.

What about Linux? Articles on Macs and Windows never mention Linux. Likewise, articles on Linux and Windows never mention Macs. Readers need some perspective.

Windows on a Mac was possible way back, before the switch by Apple to Intel processors. The company that actually developed what is now known as Virtual PC was Connectix. Microsoft bought them out and wasn't very interested in software that ran on Apple computers.

More bad advice from Walter Mossberg

In the March 8, 2007 Mossberg Mailbox column in the Wall Street Journal, a reader asked how he could make sure that no one could piggyback on his WiFi wireless connection. Walter's response was:

"Turn on the password feature in your router, and don't tell anyone the password. You'll usually find the password setting in the installation software that came with the router."

Not quite.

When discussing wireless routers, there are two passwords. I can just imagine the poor reader of the newspaper changing the wrong password and thinking he is safe. False security is worse than no security.

The first password is needed to login to the router itself. Routers have internal websites that you use to make configuration changes, and access to the internal website requires a userid and password. This password has nothing to do with WiFi wireless signals/connections.

By the way, this password should be changed when a new router is installed because all the bad guys know the default passwords. I have an earlier posting on this blog about how important it is to change this router password.

The password that prevents someone from piggybacking on your wireless connection is referred to in all the technical literature as a "key". If the reader looks around the internal router website or the router documentation for a "password", he won't find this. All references to "passwords" refer to the first password, not to the key.

Not to get too technical, but this "key" relates to an encryption standard, either WEP (bad) or WPA (good) or WPA2 (good). And there are good keys and bad ones, an important concept omitted from the response.

So, which password was Mr. Mossberg referring to? Did he have the right concept and use the wrong term, or did he have the wrong concept and use the correct term? Beats me. The PC industry is too new to have a concept of malpractice, but if the shoe fits...

---

Update: On March 15, 2007 Mr. Mossberg issued a clarification. See Securing a Wireless Network. Certainly a step in the right direction, but...

Quoting: "To enable the encryption key, use the router's setup software to turn on security".

There are multiple mistakes in this sentence.

You enable encryption, you do not enable the encryption key. Encryption is the lock, without the lock, having a combination does nothing.

This ignores the fact that the older type of encryption, WEP, has multiple keys/passwords. Only the newer type (WPA) has a single key/password.

Once the router is working, there is no need to use its setup software. Instead you log in to the internal website in the router to make changes.

You don't "turn on security". What he meant to say was you turn on one of the three types of encryption. Routers have multiple types of security, a point he makes later. Mac address filtering, for example, is a security feature having nothing to do with encryption. So too, disabling remote administration, turning off UPnP and not broadcasting the SSID.

Quoting again: "On newer models, the strongest security system is called WPA..."

WPA is not a security system, it is an encryption system (see above). And, the strongest system is actually WPA2, not WPA.

There is a big omission here: for WPA and WPA2, the length of the key/password is critical. Short is bad, long is good. To encourage long keys/passwords you sometimes see references to a "pass phrase". This means the key/password can be an entire sentence. And it should be. A very long key/password is not an ongoing typing annoyance because it only has to be entered once on each computer that want to access the WiFi network. If your WPA key/password is "dog" or "rose" you have no defense at all from a determined bad guy.

And, WPA is not a single thing. There are multiple types of WPA and the people Mr. Mossberg claims to write for need to be told this. Simply put, home users want WPA-PSK. This is also called WPA Pre-Shared Key and WPA Personal. They do not want WPA with a RADIUS server.

Finally, anyone serious about WiFi security should turn off the WiFi network when not in use. There is an option in the router to turn off the radio transmitter that is the wireless network. No hacker can break into a network that doesn't exist.

Mr. Mossberg is loose or sloppy with words/terminology. When nerds are talking amongst themselves, they can be sloppy with terminology because they all understand anyway. But when writing for a non-technical audience, it is important to be very precise when describing something technical because they don't know the ropes.

And there is no excuse for making technical mistakes, something the editors at the Journal share the blame for. Apparently no one reviews his work.

Dumb Article on Dumb Terminals

There was an article today in the Wall Street Journal called Dumb Terminals Can Be a Smart Move (this is one of the very few articles that the paper makes available for free on their web site). As is typical of most computer articles in newspapers the technical information is, to be kind, incomplete.

For starters, this long article fails to mention Citrix, for a long time the dominant player in this field. Citrix is to thin client computing as Xerox is to copiers. Also omitted is any mention of the server versions of Windows and the specific feature (terminal services) that enables thin client computing. It is impossible to give the reader a reasonable understanding of the subject of thin client computing without some discussion of the server side impact.

The article gives the impression that dumb terminals are cheaper than PCs which is not always true. Back a few years ago when I was more familiar with this subject, many dumb terminals were more expensive than new low end computers. This is still true today. The HP model pictured with the article is $200 - sort of. HP says this price is "applicable to the lowest-price configuration for this model; prices for this model with the features displayed here will vary." Wyse doesn't list prices on their web site and only sells through resellers - a sure sign that it's expensive. Neoware is up-front about their prices. Their Linux based models range from $260 to $1,100, their WindowsCE models range from $360 to $1,050. They also sell Windows XP embedded models for $480 to $1,150 and thin client notebooks start at $750, roughly the same starting price as real notebook computers.

Much of the article is about how "dumb terminals" are cheaper, but it avoids the server side costs, which are huge. The software to run applications remotely is expensive and complicated to set up. The techies in charge need a lot of training to get up to speed (a hidden cost). And then there is the hardware cost for very high end servers.

There are downsides to thin client computing that are not mentioned in the article.

For one, network access to a server running your applications is critical. Should there be a networking problem, employees do the crossword puzzle until it's fixed. Also, just like with mainframes, many users compete for resources on a single computer. If the machine is not configured right, or suffers a problem or is hogged by one user doing something resource intensive, multiple users feel the pain. In fact, the whole idea of thin clients is a throwback to the mainframe model of computing - with both its pros and cons.

Another downside, user psychology, is mentioned in the article, but there is more to say about it. Resistance to thin clients comes from users that can no longer install software. Instead they can be limited to running just approved corporate applications. This sounds great for security and for keeping software up to date (think Windows Update) but if a user wants or needs other applications, they have to fight with the computer techies to get approval. It's easy to see how this may not go over well.

Artists are another problem not mentioned. Thin clients are not appropriate for people dealing with photographs, videos or any type of graphic work. These applications require high resolution, lots of colors, fast response time and lots of cpu horsepower.

The article quotes a Gartner employee as saying: "It's a paradox. People want their cake and eat it, too. They want the security, they want the consistency ... but they want the functionality of a PC." There is no paradox at all, the techies want security and consistency while end users want normal PC functionality.

Quoting the article: "... the basic terminals appear to offer improved security. Because the systems are designed to keep data on a server, sensitive information isn't lost if a terminal gets lost, stolen or damaged."

While true, this is far from the whole story. Thin client devices have ports for connecting all sorts of external devices that can be used to save files. As with a real PC, a person can stick a thumb drive into it and copy all manner of files.

The article gives the impression that the choice is either standard personal computers or thin clients/dumb terminals. This couldn't be more wrong. Almost every Fortune 500 company uses software from Citrix to run remote applications. As described here, the application is installed on a server computer running special, expensive software and remote users with real PCs access these application servers over the corporate LAN or the Internet. In fact, I'm pretty sure that the Wall Street Journal itself uses Citrix software in this way.

Some applications run normally on a personal computer while others are run remotely on application servers. The remote applications are run using software from Citrix that is either installed on a personal computer or accessed via a web page. This is a hybrid approach and one that is much more common than exclusive use of thin clients.

I have no idea what this sentence in the article means: "H-P, of Palo Alto, Calif., last week announced that a line of its slimmed-down PCs, which reside in the data center and enable users to connect through a thin client or other devices, was being made available in Europe, Canada and China for the first time." Nothing about thin clients/dumb terminals has to with PCs residing in a data center, it's all about high end servers in the data center. And if the PCs are in a data center, who cares about their height and width?

The article discusses a thin client laptop that connects wirelessly to "the network". Which network? The LAN of a company or the Internet? It doesn't say. And which wireless; WiFi or a wide area data network such as EV-DO or WiMax or HSDPA or UMTS? It doesn't say.

Whew.

By the way, you can get a feel for running remote applications at windowsvistatestdrive.com a site that Microsoft set up to let people try/test drive the Vista operating system. I found it slow, but your mileage may differ.