Random("Technical Musings At".Length)

Android for $35/mo

2012-01-24T16:57:00.004-05:00

In the market for a smartphone. Would love to get the Samsung SII but hate contracts and high monthly service fees. Virgin Mobile has very attractive pricing. Minimum plan for their smartphones was $35/month (300 min, unlimited everything else). One of the phones they sell is the Motorola Triumph. While this is not as spiffy as the newer phones, it is not bad. 1GHz processor but only 500MB RAM. Half the RAM of the 1GB better phones can make the user experience not quite as good. The stock phone isn't really heavily loaded with software typical to other wireless providers which is nice. Strangely, you don't even get MotoBlur user interface (not that many people would actually miss it!). But, loading up CyanogenMod on the phone would probably help a lot (and get wireless tethering too!).

Long story short; it works and is not bad at all. I loved the Verizon Droid Incredible I had before this but I'm not at all missing that phone when compared to this one.

Here's how I did it on the Motorola Triumph MX435. Different models may exist? I saw some directions/videos that where not able to be reproduced on my phone. This is the streamlined version which assumes you mostly know how to work with Android devices:

Note Virgin Mobile version of Motorola Triumph
Get Gingerbreak apk file
Get ClockworkMod version 5.0.2.6 (newer versions did NOT work for me).
Get CyanogenMod version 7.1.0-TG-Beta-0.8 (I think this is final version from CyanogenMod community as they've moved on to ICS)
Get Google Apps if you want it
Install GingerBreak. Root device with GingerBreak. Verify root access with super user app. Steps for Rooting? Here ya go:

Enable USB debugging on your device (Settings -> Applications -> Development).
Connect phone via USB as external drive, open it and transfer the GingerBreak APK file to your SD card.
Now, disconnect and get a file manager from market if don't already have one installed.
Go to Settings -> Applications and check the Unknown sources option.
Now open up the file manager and browse to the GingerBreak APK.
Install it.
Now make sure USB debugging is enabled and your SD card is mounted.
Open GingerBreak and choose the Root Device option.
Once it is complete, reboot your phone and you should have SuperUser app.

To load image files, hold volume down+on/off button while phone turns on, copy image files to images folder via USB. You will have to remove battery to reboot phone after file copied. Backup stock recovery.img file. Copy over ClockworkMod recovery.img file.
To load recovery.img file (now the ClockworkMod recovery.img file), hold volume down+up+on/off button while phone turns on for at least 8 seconds. Volume up/down moves selection, menu button selects, back arrow goes back a menu. Backup your existing ROM. Factory reset phone. Install the CyanogenMod zip file. Install Google Apps zip if you want it.

If you don't know what's going on here or have trouble remembering, here is the summary. The phone can boot image files. By default it boots the default img file all on its own. By holding volume down+up+on/off button for eight seconds during boot up, it will load the recovery image file (which you've hopefully replaced with clockworkmod file). When booted into the recovery image, you can install ROM zip files (like CyanogenMod).

If you have trouble finding all the files above from the creator source (Google is your friend), you can get them here if you trust me (which you shouldn't!). I think I was able to get all from the creator source except the early version of ClockworkMod. I think? Can't remember exactly. Means my phone could be spying on me! Oh well, no banking from the phone.

Other sources of info: XDA-Developers, Moto Triumph section

Lithium Ion batteries

2011-02-18T13:14:00.008-05:00

Laptop batteries have less than desirable lifetimes. Starting to see this very clearly on the warranty prices offered on batteries.

Dell E5500 series 6 cell battery is about $150. The 9 cell is about $180. The default warranty on their batteries is one year.

Three year warranty on the 9 cell at time of laptop purchase is approximately $105. $105 is about 58% of the cost of one battery. So, if a hundred of these laptops were purchased with the three year warranty, their price indicates they expect about 58 of the 100 batteries to be turned in for warranty replacement within that three years. (Assuming costs stay same, and desired profit on warranty is similar to goods sold, and their expected return rate on warranty items... whatever that may be). Still, 58/100 is pretty high for three years?

It gets a little more interesting when looking at the other warranty offers that come later on in the laptop purchase under the "accessories" (maybe these are third party warranty contracts offered through dell?). And the interest isn't solely in that you can get the same equipment covered by two warranties (maybe three with complete care?). These are apparently for both the 6 and 9 cell versions. Price doesn't change. The 2nd year of battery life warranty is about $55. The 2nd AND 3rd year of battery life is about $62. You can probably see where I'm going with this. The second warranty offer is much less and if they fail, they seem to fail early.

Year 2: 55/150=37% (6 cell) 55/180=31% (9 cell).
Year 2,3: 62/150=41% (6 cell) 62/180=34% (9 cell).

Of the batteries that have made it through year one, they expect replacements of 37% on 6 cells on 31% on 9 cells (same assumptions as above). Notice that you can't buy year 3 alone. Years 2 and 3 are inclusive. So, the percentage only increases 4% and 3% respectively. There are two ways to look at this, either batteries that have lasted this long are very reliable; or, almost all batteries have already failed by this point. Having been witness to the many battery laptop failures, I'm thinking the latter.

Assuming the people who calculate these warranty prices use same models, you can combine these three offers to make some determinations about the 9 cell.

Year 1: 24% = 58% - 34% (3 year offer - years 2 and 3 offer)
Year 2: 31% (Year 2 offer)
Year 3: 3% = 34% - 31% (Year 2,3 offer - Year 2 offer)

As always, I'm sure there's more to these calculations than we see. I'd bet people seeking warranty coverage probably drops significantly as time elapses. And, the cost of the battery production probably decreases over time as well.

No matter how you look at it, the cost of the warranty options shows they are anything but long term reliable. Considering the new electric cars are using similar battery technology how does this fare for their warranty?

"You can't give 'em away!"

2010-09-14T17:40:00.004-04:00

NewEgg practicing the quote (no option to remove the free item). I'd bet the reason for this is they would spend more money trying to "recycle" the e-waste than they would shipping them to people like me?

But, this does bring up a nifty idea! When ever I ship something in the future, I can just include a little bit of my garbage! Save money on garbage collection services in the office. Recipient: "Why look! Here's that new laptop from the company. And what's this? A banana peel, a plastic cup, and some wadded up paper?" Okay, maybe not.

Android - 90 days later

2010-07-16T18:15:00.009-04:00

When asked about the iPhone (4) and Android (2.1) pro's and cons. Briefly, here are some I see as core comparison items in my opinion after using it for three months. This is by no means exhaustive. Just the major points that stand out from my use.

Android Pros:

Phone - Press "Phone" and dial or type the letters of the contact. It is a phone after all, it should efficiently dial just like all the other phones. You can also drop an icon shortcut of any contact on any of your screens as a single touch call button.
Widgets - Customizable display elements of all sorts.
Predictive Keyboard - It automatically picks the most likely word as you type based upon the proximity of the letters near where you pressed. Some come with Swype by default. Others you can get it as an app without rooting the phone.
Google Search - Includes voice recognition.

Android Cons:

No ability to sort contacts by company (out of the box anyway).
International dialer does not convert "+" to international out prefix (on Incredible and Droid X thus far). With global contact lists for people used in different countries, this is a real PITA.

iPhone Pros:

Video calling. (To other iPhone 4 and 3G or better network for both parties). Maybe on Sprint Evo and Droid X as well as app incarnations?

iPhone Cons:

Dialing by picking a contact from a list is tiresome affair.
No sorting contacts by company either.
Lockstep uniform interface adhearance. Some like this and some despise the loss of design freedom. No widgets or swype keyboards for you!

Blackberry still seems to have the best keyboard for pounding out text from a phone. Hardcore Blackberry users can still double my best touchscreen keyboard time for typing sentences.

Hardware wise, some advantages go either way depending upon manufacturer. Some Android 2.1 devices have 8 MP cameras and can record HD video. Antennagate might ruin a call or two on iPhone 4 for people not holding it in that certain way.

Run explorer as admin

2010-06-18T13:09:00.006-04:00

One of those annoying troubles with microsoft windows machines has been inability to connect to file shares on a single file server using different credentials. Using the runas command with explorer.exe doesn't quite work like it does with the other programs. Trying to run this command:

runas /user:domain\administrator explorer.exe

One would get an error message that says:

[Window Title]
explorer.exe
[Content]
Windows cannot access the specified device, path, or file. You may not have the appropriate permissions to access the item.
[OK]

There is a workaround according to MS. Running the Windows Explorer Shell in Administrative Security Context While logged on as a normal user:

Start Task Manager. Right-click the Task bar and select Task Manager.
Click the Processes tab.
Select explorer.exe. Click End Process. Click YES on the warning pop-up message. The entire desktop will disappear. You will still have any programs that you started including Task Manager.
Click the Programs tab.
Click New Task.
Type runas /user:machine/domain name\administrator explorer.exe. Click OK.
A Console window will appear and prompt for the password. Minimize Task Manager, type the password and press Enter. The desktop will return including the task bar, shortcuts, Startup folder items, etc.
Perform necessary administrative tasks. For example: clicking Start, Settings and Control Panel will bring up control panel in administrative context.
When finished, log off Administrator. A new shell will automatically start, running in the originating user context.

(copied from http://support.microsoft.com/kb/225035)

Use any copier/printer

2010-05-24T17:28:00.008-04:00

Quick trick for those who find they need to print to devices while traveling. Often, copiers and network printers are accessible via traditional means through a print server. The print server is often secured. Depending upon the company, it might take a few minutes to days to get security clearance to the print server. Rather than going through the print server, we can send print jobs directly to any printer using one print driver. The following is true and tested on Windows 7 and Server 2008. Directions are different but similar for older versions of Windows.

You'll need the printer IP. You can usually get this from the printer itself via the screen or by printing a diagnostic page. This might be the most difficult part. Methods vary by make so you'll have to look around. Alternatively, you can get it from somebody nearby who usually prints to it. Devices and Printers --> Right Click printer --> Printer properties --> Ports tab --> Port column. Usually, the port name is the same as the IP but not always so you can click "Configure Port..." button to verify.

Once you have the printer IP, click "Add a printer" --> "Add a local or network printer" --> Pick "Create a new port" --> Select "Standard TCP/IP Port" --> Enter IP address --> Helps to name the port "ANYIP" as it will change for different equipment --> UNCHECK "Query the printer and [blah blah]". At this point, the computer may attempt to automatically identify the device at the other end anyway. It will probably pick a "Generic Network Card" which is fine. Click "Next" button and pick "HP LaserJet 5". If that driver doesn't exist, pick "HP LaserJet 4100 PCL6" (just about any of the old generic PCL 6 drivers should work). Name the printer whatever you wish. I append "(Print to IP)" at the end to make it easy to find. Print test page to verify setup worked.

If you travel to a different office, now you can "Configure Port..." in steps mentioned to get IP to set the port to a new IP.

There is a "gotcha" to this. Almost all of the printer settings are useless as they are specific to the printer driver you picked and not the actual device on the other end. Because of this, try to setup the page as much as possible through the program doing the printing rather than the printer properties.

Office 2007 Outlook Search

2010-03-17T21:31:00.012-04:00

The newer Office 2007 Outlook search appears to use a different indexing mechanism than the old Office 2003 Outlook search indexing mechanism. If you run 2003 indexer, the old search methods are still there via Shift+Cntrl+F. The right click search option is gone, however. The 2007 searches are said to be faster which I found to be true. This was mostly apparent for searches across large archives.

New search is started from "All Outlook Items" or the search input box above the table of emails. On key difference from new to old was the old had fairly well laid out GUI for selecting which folders to search. The 2007 search is not so clear. The drop down box on "All Outlook Items" allows root level control of where the search is performed. The drop down box on the search input box is for that currently selected folder and it also appears to have the root selection option, however it requires a restart of outlook to take affect (read closely... one affects search folders the other affects index locations and thus, the restart). You can also pick to recurse subfolders in these options. It might seem to have less control of the folders but it is actually better in a few aspects. The input box allows you to input "folder:thisone AND NOT folder:notthisone". This was not easily done in 2003. Of course, some will say learning search syntax is not as intuitive as the GUI which is true. After using it a few times, it does become fairly easy to get familiar with it and I've found I can construct searches faster than I did in the previous advanced version. You can even group together items for good logical searches like, "(to:myname OR cc:myname OR from:myname) AND sent:today" (all my email activity today).

. In 2003 search was nice in that it displayed the results in a seperate window and you could do other things while the search finished. The 2007 search is a bit less obvious about holding the results in a seperate window. If you right click either the "All Outlook Items" or the search bar, you get a chance to open a new window to do the search. This can also be done after the search is run and it will preserve the results. If you accidentally navigate away, you can use the back button on the advanced toolbar to go back to the search window but the results will be wiped out. Redo the search from the drop down and select "Recent Searches". As of this writing, there does NOT appear to be a way to save these searches as a search folder. (Booo!)

Get all the search syntax you ever wanted to know here: http://www.microsoft.com/windows/products/winfamily/desktopsearch/technicalresources/advquery.mspx. Additional sources here: http://searchexchange.techtarget.com/generic/0,295582,sid43_gci1357566,00.html.

Correct activation key error in Office 2007

2010-03-01T09:48:00.006-05:00

If you accidentally enter an old key in office 2007 activation wizard, it will keep it and not activate and NOT allow you to change it. The MS recommended fix is to uninstall office and reinstall it. (Brilliant; I know.) Having worked with interconnected office products, I know this can cause major problems later on. For instance if you already have Visio or One Note installed and remove the core office install, what happens? Don't know if the newer versions are less buggy, but older versions this was not really planned for.

So, the non-MS recommended (better?) solution is to delete the keys that are holding the wrong activation info. They are here (not exact match as guid may be slightly different):

(32 bit OS)
HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Office\12.0\Registration\{30120000-00NN-0000-0000-0000000FF1CE}

(64 bit OS)
HKEY_LOCAL_MACHINE\SOFTWARE\Wow6432Node\Microsoft\Office\12.0\Registration\{30120000-00NN-0000-0000-0000000FF1CE}

If you have more than one that looks similar look in them for a key named "Product Name" to id the one which needs to have its activation corrected. Delete these two keys:

DigitalProductID

ProductID

Open office and reactivate with valid key.

Sources:
http://support.microsoft.com/kb/919895
http://www.mydigitallife.info/2007/01/12/how-to-reset-or-change-microsoft-office-2007-product-license-key-or-volume-license-key-vlk/

As with anything involving registry changes, make these changes at your own peril

SQL run job if not running

2010-02-17T22:37:00.004-05:00

-- 1. Verify the job is not already running. If job not inactive, display message "Job not ready to run".
-- 2. If job inactive, run the job. Provide email if desire notification when complete.

use msdb
-- Setup ability to select from stored proc.
if exists (select * from master..sysservers where srvname = 'loopback')
    exec sp_dropserver 'loopback'
go
exec sp_addlinkedserver @server = N'loopback',
    @srvproduct = N'',
    @provider = N'SQLOLEDB', 
    @datasrc = @@servername
go

-- Run the mainjob if inactive.
if (4=(select current_execution_status from openquery(loopback, 'msdb..sp_help_job @enabled=1') where [name]='MAINJOB'))
begin
        exec msdb..sp_start_job @job_name='MAINJOB'
        waitfor delay '00:00:05'
        while 4<>(select current_execution_status from openquery(loopback, 'msdb..sp_help_job @enabled=1') where [name]='MAINJOB')
        begin
                waitfor delay '00:00:05'
        end
        exec sp_send_dbmail 
                @recipients='user@domain.ext'
                ,@subject='MAINJOB initiated via web. Job complete.'
                ,@body='Your email address was entered to receive notification when this sql job completed. The job is complete.'
end

-- Cleanup
if exists (select * from master..sysservers where srvname = 'loopback')
    exec sp_dropserver 'loopback'
go

select from stored procedure

2010-02-17T22:33:00.002-05:00

use msdb
-- Setup ability to select from stored proc.
if exists (select * from master..sysservers where srvname = 'loopback')
    exec sp_dropserver 'loopback'
go
exec sp_addlinkedserver @server = N'loopback',
    @srvproduct = N'',
    @provider = N'SQLOLEDB', 
    @datasrc = @@servername
go

select current_execution_status from openquery(loopback, 'msdb..sp_help_job @enabled=1') where [name]='JOBNAME'))

-- Cleanup
if exists (select * from master..sysservers where srvname = 'loopback')
    exec sp_dropserver 'loopback'
go

Price per MB as of 2/17/2010

2010-02-17T22:18:00.004-05:00

Quick price list from various websites for wireless internet service.

Virgin (prepaid) (sprint network?)
10/100  = $0.10/MB (10days)
20/250  = $0.08/MB (30days)
40/600  = $0.07/MB (30days)
60/1024 = $0.06/MB (30days)

Verizon (prepaid)
15/75  = $0.20/MB (1day)
30/250 = $0.12/MB (1wk)
50/500 = $0.10/MB (1mo)
Verizon (24 month)
40/250  = $0.16/MB $0.10/MBOvr
60/5120 = $0.01/MB $0.05/MBOvr

Sprint (24 month)
60/5120 = $0.01/MB $0.05/MBOvr (4G is unlimited)

ATT (24 month)
35/200  = $0.18/MB $0.10/MBOvr
60/5120 = $0.01/MB $0.05/MBOvr

TMobile (24 month)
30/200 = $0.15/MB
60/5120 = $0.01/MB

Slingshot (prepaid) (vaporware?)
60/5120 = $0.01/MB
50/500  = $0.10/MB
35/250  = $0.14/MB
25/100  = $0.25/MB

Flash ads shoot themselves in the foot?

2009-06-19T08:57:00.003-04:00

I know this may sound crazy, but every once in a while, I do see an ad online I actually am interested in learning more about. But, here's the problem, many ads will take me away from my current page (not desirable) or they want to open a new window (pop-up blocker prevents). Turning on/off the pop-up blocker causes the entire page to refresh so you lose the ad. Effectively, the ad has to be MORE interesting than the current website I desired to view. (Not an easy task).

Toggle the Blogger Navigation Menu

2009-06-07T09:16:00.009-04:00

Since using Blogger, one think I've always wanted was the ability to widen the content area by minimizing the navigation menu. After a few hours, I got it working. You can give it a try over to your left there. (Click the «) You can freely use if you wish. Just make sure to keep my credits in the source. ;-)

You will need to edit the template HTML in several places.

First, modify the style sheet area for the navigation area to look something like this:

#sidebar-wrapper {
width: 220px;
float: left;
/* word-wrap: break-word; /* fix for long text breaking sidebar float in IE */
white-space: nowrap;
overflow: hidden; /* fix for long non-text content breaking IE sidebar float */
}

Next, you will have to search through the template a bit towards the end. If you haven't already, save and make sure you have "Expand widget templates" checked or else you won't see the section. The section you are looking for will look something like this.

<b:includable id='main'>
  <b:if cond='data:title'>
    <h2><data:title/></h2>

Change it to look like this modified snippet.

<b:includable id='main'>
  <b:if cond='data:title'>
    <h2 onclick='NavDebug();'><data:title/></h2>
    <div style='text-align:right;'><a id='aNaver' onclick='TogNavMenu();' style='cursor:pointer;'>«</a></div>

Finally, you will need to add a javascript block to the template. This is the javascript code.

<script language="JavaScript">
<!--
/*
 * NAME         : BloggerToggleNav
 * VERSION      : 1.1
 * AUTHOR       : Adam Griffin (google.adamgriffin.net)
 * DONE Upgrades:
 * ~ Cookie remembers nav state.
 * TODO Upgrades:
 * ~ Error changes nav char to "Err" and error code in div that appears on mouse over.
 * ~ Verify script works with FireFox.
 */
var tNavMenu;                // Timer used for sliding div width in and out.
var iNavMenuSlide = 10;      // The amount of pixels covered in slide.
var iNavMenuSlideSpeed = 10; // The milliseconds between slide actions.
var iNavMenuWidthMin = 20;   // The minimum width of the nav menu.
var iNavMenuWidthMax;        // Set on initialize fn.
var iMainWidthMax;           // Set on initialize fn.
var vNaver = "Naver";        // The cookie name used to store the nav menu state.
var vNavCharOut = "«";       // Could also use character indicators if unicode supported. Others: ◄ ▬ ► « » ˂ ˃ ← →
var vNavCharIn  = "»";
var vNavIndicatorID = "aNaver";
var vNavMenuID = "sidebar-wrapper";
var vMain = "main-wrapper";
var oNavIndicator;
var oNavMenu;
var oMain;

var bInitialized = false;
function Initialize(){
  try{
  
    // Set initial values.
    if(oNavIndicator==null) oNavIndicator = document.getElementById(vNavIndicatorID);
    if(oNavMenu==null) oNavMenu = document.getElementById(vNavMenuID);
    if(oMain==null) oMain = document.getElementById(vMain);
    iNavMenuWidthMax = oNavMenu.offsetWidth;
    iMainWidthMax = oMain.offsetWidth;
 
 // Make sure the nav never gets in a state where it can not be seen.
 if(1>GetCookie(vNaver)) SetCookie(vNaver,iNavMenuWidthMax);
 else oNavMenu.style.pixelWidth = GetCookie(vNaver);
 
 // What state was the nav menu last in?
 if(iNavMenuWidthMax==oNavMenu.offsetWidth){
   // Nav Menu is displayed.
   oNavIndicator.innerHTML = vNavCharOut;
 }else{
   // Nav menu is hidden. Set sizes and toggle directional elements.
   iNavMenuSlide *= -1;
   oNavIndicator.innerHTML = vNavCharIn;
   oMain.style.pixelWidth = oMain.offsetWidth + iNavMenuWidthMax - iNavMenuWidthMin;
 }
    bInitialized = true;
  }catch(err){
    alert("Function:Initialize(). Error:" + err.description);
  }
} 
// Event handler when user clicks Nav element.
function TogNavMenu(){
  try{
    if(!bInitialized) Initialize();
    // Toggle the pixel slide direction.
    iNavMenuSlide *= -1;
    // What direction is it currently facing?
    if(oNavIndicator.innerHTML==vNavCharOut){
      // Was out so head in.
      tNavMenu = window.setInterval("SizeNavMenu()",iNavMenuSlideSpeed);
    }else{
      // Was in so head out.
      tNavMenu = window.setInterval("SizeNavMenu()",iNavMenuSlideSpeed);
    }
  }catch(err){
    alert("Function:TogNavMenu(). Error:" + err.description);
  }
}
function SizeNavMenu(){
  if(iNavMenuSlide < 0){
    // Is the inner limit reached?
    if((oNavMenu.offsetWidth + iNavMenuSlide) < iNavMenuWidthMin){
      window.clearInterval(tNavMenu);
      oNavMenu.style.pixelWidth = iNavMenuWidthMin;
      oNavIndicator.innerHTML = vNavCharIn;
      SetCookie(vNaver,iNavMenuWidthMin);
    }else{
      oNavMenu.style.pixelWidth = (oNavMenu.offsetWidth + iNavMenuSlide);
      oMain.style.pixelWidth = (oMain.offsetWidth - iNavMenuSlide);
    }
  }else{
    // Is the outter limit reached?
    if((oNavMenu.offsetWidth + iNavMenuSlide) > iNavMenuWidthMax){
      window.clearInterval(tNavMenu);
      oNavMenu.style.pixelWidth = iNavMenuWidthMax;
      oMain.style.pixelWidth = iMainWidthMax;
      oNavIndicator.innerHTML = vNavCharOut;
      SetCookie(vNaver,iNavMenuWidthMax);
    }else{
      oNavMenu.style.pixelWidth = (oNavMenu.offsetWidth + iNavMenuSlide);
      oMain.style.pixelWidth = (oMain.offsetWidth - iNavMenuSlide);
    }
  }
}
function NavDebug(){
  var v = "pixelWidth:" + oNavMenu.style.pixelWidth + "\n" +
          "offsetWidth:" + oNavMenu.offsetWidth;
  alert(v);
}

function SetCookie(sName,sValue){
  var d = new Date();
  d.setFullYear(d.getFullYear()+1);
  document.cookie = (sName + "=" + sValue + "; expires=" + d.toGMTString() + "; path=/" );
}
function GetCookie(sName){
  var a = document.cookie.split("; ");
  var aTmp;
  for(var i=0; i<a.length; i++){
    aTmp = a[i].split("=");
    if(aTmp[0]==sName) return(aTmp[1]);
  }
  return(null);
}

Initialize();

-->
</script>

Now you can show and hide the navigation menu on blogger as you like!

Analytics and Office Live Small Business

2009-06-06T22:17:00.006-04:00

Discovered an interesting bug while testing Microsoft Office Live Small Business. I tried pasting the analytics script into an HTML object on the web page only to discover that it causes IE8 to hang. Hmmm... Very fishy. Guess I'll have to use Microsoft's version of Google's analytics until that can be sorted out.

Also strange side note. Why does Blogger spell check correctly suggest for "microsofts" but not "googles"?

In comparing Google Sites to Microsoft Office Live Small Business Website they both offer a good bit of functionality considering they are free. Microsoft allows for a full domain to pointed to an IP while google only allows for a sub domain to be directed to their site (AName DNS record vs CName DNS record). One other advantage Microsoft appears to offer is the ability to download/upload an entire site. This definitely makes the customization factor much more weighted in its favor.

Security on the WWW (Wild Wild West)

2009-05-29T21:43:00.004-04:00

Recent security related requests at work have led me to muddle over some ways to address some security problems. So much of the security world seems to focus on the aspects that are not the main source of security "holes". It seems most security compromises are not really achieved through brute forcing public-private key encryption or passwords. There are far easier routes to get into a system. While it appears readily obvious that there is no 100% guaranteed security (even if you were to stop using computers and go back to paper!), there are some obvious measures that can be taken to minimize easy avenues of security compromises.

The best way to understand the security holes is to wear the hat of a person who exploits them. Donning the black hat, how would I get in a company? Given the sheer number of malware compromised machines at any given location, this is by far the easiest way "in" an existing system. Since this option historically works well, it would probably continue to be a highly targeted area.

Typical avenues of approach: 1. Send emails with executable attachment in zip file using known email addresses obtained from phone calls to sales people. 2. Send cheap flash drives containing "promotional material" to execute upon autorun. 3. Walk in the company and stick a flash drive into an unattended machine. 4. Email links to malware dispensing websites using typical baiting (money, pharmaceutical meds, pornography).

Of these methods, writing a program to copy another program to a running location is ridiculously easy (1-3). Writing a website to take advantage of browser exploits (4) would definitely take more time but have much higher delivery capacity to a wider audience.

With delivery complete, what kind of malware would work well? I've always wondered how hard it is to engineer a key logger. The answer; roughly four hours. And, I'm sure some have done it in far less time. In addition, that was completed in an unfamiliar development environment for me. Without tipping too much of the details out, that four hours produced a rather robust key logger and screen logger to boot. In addition it runs with no visual indicators that it is running and in a manner that does not completely hamper system performance which would help hide the fact a machine has been compromised. To keep the program neutered, no mechanism was made to get the logged data. However, getting the logged data out could be as simple as mailing it back out via SMTP or picking up USB flash drives left on computers. Programmaticly, these are not large obstacles.

All your systems belong to us! There you have it, keys of the castle delivered, the system is yours. Depending upon what kind of data you've logged, access can be obtained to bank accounts, credit cards, etc. What happens after that is beyond the technical scope and I haven't the faintest idea how to cover the money trail anyhow!

Now that several routes in have been analyzed, we can don the white hat to look at counter measures. Firewalls, both network and application are definitely a must. I don't think anybody really operates without them (willingly anyway). Antispam/malware gateways for all email traffic can help eliminate most routes in via that mechanism. Following up on that, be sure to have the email server only accept email from the the gateway IPs (ignore all others as targeted spam). Likewise, handle the outbound traffic with similar skepticism. Is there any reason for SMTP to be allowed everywhere on the network? Typically SMTP can be restricted to a limited amount of IPs so that outbound routes that malware typically utilizes are not available. In addition, route the outbound traffic back through the gateway as a further attempt to help eliminate outbound spam. Always run users as true user accounts to prevent unsecured installations or system changes. To be debated, perhaps later, teach the users how to avoid getting compromised. Experience indicates most users will not retain these lessons.

Even with all these measures. We've only stopped email and website routes for malware delivery and their logging feedback (and even then, it is still not 100%). Entry points 1-3 could still be utilized with some degree of success. Disable flash drive access on USB ports? Probably not realistic. Remove execute permissions from all flash drives? Maybe possible if not FAT file systems? So, essentially we are faced with the possibility of working on a compromised machine.

How do we minimize what can be gathered from that machine? To avoid key logging, we can use a keyboard on the screen. But, that could still be captured via screen logging. How do you defeat screen logging? You could display multiple pointer inputs and use a very brief indicator to lead to the correct pointer. The brevity would mean the screen logging would have to be essentially movie quality recording everything. This would require much more computer resources and lots of bandwidth to deliver what was logged back to the black hat. These two things would be much more readily noticed. Does such a program exist? I don't think so. How long would it take to make it? Not sure. Perhaps I will give an attempt at that similar to the attempt a making the key logger. Keep an eye out for an update.

Mobile RAID 1 2.5" Enclosure updates

2009-04-18T23:50:00.000-04:00

Updated prior blog entries to reflect usage experience of ACard product and dual port usb cord. /2008/02/mobile-raid-1-25-disks.html

Mobile RAID 1 2.5" Enclosure updates

2009-03-12T20:46:00.000-04:00

Updated prior blog entries to reflect usage experience of ACard product. /2008/02/mobile-raid-1-25-disks.html

Mobile RAID 1 2.5" Enclosure updates

2009-02-14T22:11:00.003-05:00

Seems to be more market offerings on the topic. Update prior blog entries to reflect some recent research. /2008/02/mobile-raid-1-25-disks.html

Umm... Why? (Ridiculous Requirement)

2008-09-26T17:57:00.004-04:00

I don't think this is a real "requirement" by definition. Even if they're thinking the person will be working outdoors more frequently, it should only be a suggestion.

Extended Warranty Why?

2008-06-02T16:20:00.005-04:00

Sometimes, those extended warranties just do not make any sense. (I still like NewEgg, though)

Annoying Desktop Alert

2008-03-12T10:34:00.002-04:00

Here is a quick tip for how to disable the desktop alert in Microsoft Outlook 2003 when it has been disabled by policy setting. It would appear grey and disabled so that you can not change checkbox to disable the "Display a New Mail Desktop Alert (default Inbox only)" setting. Change the following DWORD registry key decimal value to be zero ("0" without quotes). HKCU\Software\Policies\Microsoft\Office\11.0\Outlook\Preferences\NewmailDesktopAlerts As usual, anything involving the registry can be pretty risky for those who don't know what they're doing so make this change at your own risk.

Mobile RAID 1 2.5" Disks

2008-02-02T17:26:00.012-05:00

Why would you want mobile RAID 1 2.5" disks (or portable mirror array of notebook drives in same language of different type)? For me, I just like to make a backup of my data at home and at work to the same location and know that the backup is mirrored for safety. It's also nice to know the data your transporting has an extra bit of safety in case of something going wrong with one of the drives.

NOTE: A vast majority of this posting is about early attempts at RAID and notebook drives. Skip to the bolded update sections towards the end for more updated info.

Some shopping around revealed a few RAID 0 options for purchase but I couldn't find a single RAID 1 option using notebook drives. There are a few options out there for RAID 1 3.5" drives. One 3.5" drive is about the limit of portability for me so I wasn't looking to exceed that form factor. I happened across this company that makes RAID controllers for 3.5" form factor solutions (meaning the controller and disks should fit in a 3.5" slot). It's mainly targeted for use in a desktop/tower configuration as the complete length is about 9". But I thought maybe I could find a somewhat spacious enclosure that would hold it and the two notebook drives. For those interested, the company URL is: http://www.ezd2d.com/ . (The specific product I used is "MicroRAID").

Looking at enclosures online, I found a few with the exterior dimensions exceeding 9" and using an IDE interface (a requirement of the MicroRAID controller). I researched a few that seemed to allow for some extra length inside the enclosure and chose the SilverStone NS312. Company URL is: http://www.silverstonetek.com/ . This unit also has the added advantage of fulfilling NAS functions. Reviews seemed positive but somewhat lacking of detail.

For drives, I chose two Toshiba 100GB 7,200 RPM drives. They where cheap and just about the largest notebook IDE drives I could find with a 7,200 RPM spindle speed. I didn't choose SATA because the MicroRAID controller for SATA drives is not available as of this writing. In addition, I know these notebook drives aren't going to come close to exhausting the IDE bandwidth anyhow so there's no real advantage to SATA in this particular case. I know some people might have concerns over the Toshiba notebook drives due to past reliability issues but I figured I'd give them a try hoping for improvement from lessons learned kind of thing. In addition, I haven't been hearing anything terrible about the current models.

The setup of the MicroRAID drives and controller was pretty straight forward. The only hitch I encountered was during the initial format of the drive. I did the initial format with just one drive in and then inserted the second to test the rebuild procedure. I was expecting it to automatically start the rebuild process on boot up but it did not. It turns out the rebuild must be initiated through the software provided with the MicroRAID. This isn't a huge deal, but there are things to consider for our use here. This means if one drive fails in the enclosure, the MicroRAID controller must be removed and attached to a computer with the software installed to perform a rebuild. So, no hot swap (to be expected with IDE interface), but no cold swap either. It would have been icing on the cake.

With the MicroRAID controller tested and working, I went about examining the fit to the NS312 enclosure. With the tray out, it looked like it would fit. Fitting it to the tray proved to be impossible without alteration. The lip at the "front" of the tray had to be flattened so the holes on the side would align to the tray. A pair of pliers did the trick. The rubber mounting holes are a good idea for shock absorption. Everything "fits" although the tray holes do not align. This turns out to not be an issue as the horizontal movement of the tray is just about completely impossible due to the tightness of fit on both ends plus the pressure exerted once the cover is screwed down over the tray. The following pictures document the fit.

Tray before and after the front lip is flattened.
Tray holding the MicroRAID controller with drives.
Tray in enclosure with minimal clearance.

The NS312 ended up being the most difficult part to work with, and not for its physical capacity limitations. The first model I was shipped did not have working power to the drive so was completely unusable. The second model I was shipped was missing the software disc. This I was able to work around but these two things seemed completely uncharacteristic for SilverStone who seems to enjoy a "high end" reputation for excellent parts. Issues I experienced do not end there.

For whatever reason, certain things do not work as expected on a more complex network. The office where I work is a 10.1.x.x network with VLANs for voice and data. For whatever reason, it would not work through my switch in the office on this network. I didn't waste time troubleshooting why (maybe a switch issue, maybe a VLAN issue, maybe a 10.1.x.x issue) and resorted to a direct connection to the device via secondary network connection on PC to device. Fortunately, crossover cable was not required.

As of this writing, the manual has some fairly serious errors (both the one shipped with product and the one online). The name of the device is stated to be "DSTNAS" on one page, "DST-NAS" on another, "SSTNAS" on the online manual. Leaving names aside, using the IP listed in the shipped manual and the online manual were also incorrect. In practice it was NONE of these names nor IP addresses! Some educated guessing led me to try 192.168.1.1. Low and behold, that worked and I got an login screen for the device. I also discovered the default name being used by the device was actually "STORNAS". That none of this is listed in any documentation is blot on SilverStone's reputation (how hard is it to update your online documentation!?).

The difficulty did not end there. Using the setup wizard on the administrative web page led to some problems. The password part of the setup states that you can just click next if you do not want to change the password. I have no idea what the password was changed to but it was NOT the default. To make matters worse, the "reset" button on the device was undocumented. I tried some intuitive guesses like holding it down till it rebooted, and holding it down during reboot for at least 15 seconds. None of these reset the password. Thinking that it might be stored on the drives I hooked it to the computer via USB and wiped the three partitions listed for the device. Note the USB was not detected if plugged in while already powered up. You must have it already connected and then power the device on. So, with all partitions removed and a quick format performed, I rebooted the device and found the login and password were the default values again.

This time around, I skipped the wizard portion of the setup and did manual configuration. Weird things occurred like being unable to change the hostname more than once, and inability to disable DHCP. It turns out there are some undocumented automatic actions occurring here. If you have your device set to automatically assign an IP and it fails to find one, it appears to turn on the DHCP services. (Yet another thing that would have been nice to know). You can see how this could be a bad thing in an office to have other computers suddenly getting their leased IPs from THIS device instead of the domain controller!

After discovering all kinds of little oddities, including total loss of NAS functionality, I figured this is a device best left to minimal configuration changes. Another reformat was done and, at that point, I'm through with being beta tester.

On the simple home network, more issues occurred. I use a basic DLink router. It uses the IP range 192.168.0.x and a net mask of 255.255.255.0. To accommodate the device being 192.168.1.100 (static setting from work), I changed the mask to 255.255.0.0 and performed a "repair" on the PC LAN connection which grabs an IP and begins using a net mask of 255.255.0.0 as well. For whatever reason, I still could not find the device. I resorted to direct network wiring on second network port (statically configured) and found it sitting at 192.168.1.100. Made a change to automatically assign IP, shut down, attach to router, power on, and find it is working with an automatically assigned IP.

The network issues might prove to be a real pain if I can not find a quick and reliable method for attaching the NS312 in various locations. So far, I'm thinking USB might be the only reliable method. This would be a sad degradation of service since the nice thing about NAS functionality is being able to access the device from other computers on the network.

I should note here the device will only work as a USB drive if you use the FAT32 format option. This is not advertised but is documented in the manual. Other formats offered are Ext2, and Ext3. Network connection is as advertised to be 10/100 (1000 would be nice).

The dating for the material in this post is Feb 1st, 2008. The issues I experienced on the NS312 occurred while using firmware labeled as: Hardware Name PLN6800 Platform Firmware Version 2.3.2.SS.1 (Oct. 23,2006) Hardware Version Rev. B

Considering there is not updated documentation, nor new firmware available for download off the website, and it is now 2008, and this product still has some issues... I thinking it is an EOL (end of life) type product.

Gripes aside concerning the NS312, the physical construction is quite nice. The all aluminum construction seems fairly precise and well done. The size is appropriately mobile. It has holes on the short sides to allow for ventilation. No excessive heat build up was noticed in my use. The basic form factor and saving grace of USB connection will allow this to fulfill my intended purpose of mobile RAID 1 notebook drive configuration.

Some after thoughts: Googling "STORNAS NS312" reveals other people who have actually set this product up... only one other review: http://www.reviewdesk.com/silverstone-ns312-network-attached-storage-device/3/ .

02/19/2008 Update: Press release a few days ago caught my eye: http://www.i-newswire.com/pr151424.html. It would appear SansDigital has also been thinking along the same lines and developed a similar product that could be used in a similar fashion as I'm using above. Notable difference is their claim to full fill the exact same dimensions as a normal 3.5" hard drive. I will be trying their combo out to see if it works. InstaRAID IR12TB $185 from their site. MobileSTOR MS1UT $65 from their site.

03/11/2008 Update: I received the IR12TB but found it is probably broken. The HD's work fine individually, and in RAID0, but not in RAID1 configuration. Unit consistently indicates drive failure when in RAID1 mode. Currently in process of getting an RMA. Emailing tech support from their web page did not seem to yield a response (unless spam filtered?). Phone call did the trick (they appear to be 9-5 PST... no call center message indicated what it was). Tech support was smart and unscripted type. Hard drives used are two Western Digital Scorpio 320GB 5,400RPM SATA 2.5" drives (WD3200BEVT).

06/01/2008 Update: I received the fixed equipment a while back. It took longer than usual to get it back. They attributed this to moving offices. After much tinkering, it does not appear that one can get the IR12TB to work in the MS1UT. My best guess at this point is it appears there is a boot up time required by the IR12TB that does not work well with the MS1UT. Each part works independently well on their own. Put them together and they just do not play well. Sometimes it will work, sometimes not (not changing ANY settings; just powering it up and watching). Needless to say, the whole point of RAID is reliability so the "sometimes not" potential screams for them not to be used as such. I like the storage volume (320GB), RAID 1, and form factor of the IR12TB so I will continue to use it. However, continuous transportation, and plugging in/out of connections was probably not what it was designed for. In addition, no USB and eSATA means it is only going to work on computers with cables hanging out. It is not a big deal for my home and work computers but not good enough for use at other's computers.

Also, from the SansDigital press release department, they've release a device for RAID 1 compact flash devices in a 2.5 form factor. They've named it CompactSTOR CR2T. It is very interesting, but compact flash has not yet achieved a storage capacity suitable for my uses. (Note that they also include spanning option to help with capacity demands.) I imagine this product would be useful in some extremely ruggedized laptops. RAID 1 compact flash is about as rugged as you can get in a laptop.

02/14/2009 Update: A few months ago, I purchased a portable dual drive RAID enclosure from addonics. You can view it on their website, model AE25RDESU. On paper this device meets the requirements. In practice it has faced a few problems.

The number one problem has been heat dissipation. In my case, I'm using the two Western Digital Scorpio 320GB 5,400RPM SATA 2.5" drives (WD3200BEVT) from previous project. These will work reliably for about 20-30 minutes before they get so hot that something malfunctions. Either the drives or addonics device becomes too hot to function. At this point, the AE25RDESU is almost too hot to hold. The only way I could get the RAID to finish building the first time was by putting a glass of ice water on the thing (and some paper towl wrapped around the glass to absorb condensation).

The number two problem has been a very touchy power connector. Slight bumps to where the power plugs into the back of the AE25RDESU can trip the power off. If this happens at a bad time or during boot up of the device, it seems to corrupt the RAID. After this, the device has to rebuild the mirror (and its another trip to fridge of ice water).

These problems still leave the marketplace without much of a real solution for portable RAID enclosures for the 2.5" form factor. Amazingly, I found what appears to be the exact same product sold under a different company http://www.cooldrives.com/2usb2alrahdd.html.

More searching around on the Internet revealed a similar device of different construction. The Sonnet Fusion F2 has similar features and a few more. It also costs significantly more at $460-600 for the FUS-F2-1000M5 model depending upon vendor as of this date. That cost is far more than I am willing to pay. Especially when 500GB drives are going for $100! The product does seem targeted towards the Mac user with its slim construction and aluminum shell. They might be thinking the Mac user is also flush with cash?

In a similar form factor with vastly reduced features and vastly reduced cost is a product from ACard model ACP-2212P. Their website has a comical translation attempt that is both funny and painful to read at the same time. Definitely Engrish quality material. Unfortunately, no eSata port. But, might still be a good enough product for the original purposes.

03/12/2009 Update: I've been using the ACard model for a solid month now with out issue. Zero overheating issues. I've run it through the paces with a mirror rebuild and thrown about 300 gigs onto it in a single copy operation without any issues. The form factor isn't too bad with dimensions a little smaller than your average mouse pad. The required power supply hurts the portability factor a bit. It'd be really nice if the entire operation could be powered off two USB ports (like some other devices have done when they require a little extra juice). I might give it a try with this device or something similar.

Also ran across another portable RAID model that appears to use stacking instead of the side by side configuration. It is the G-Technology G-RAID mini2. Yes, I was surprised it wasn't named G-Technology G-RAID G-mini G-2, as well. It does seem to have a few more features than the Sonnet F2. Funny enough, this model appears to also be fairly expensive at $700 for the 1TB model and is mostly targeted at Mac users. Nothing says "money to burn" like owning a Mac apparently?

Power Over eSATA will definitely make a mark on items discussed here. (Ars Technica article)

04/18/2009 Update: Attempted the power the ACard unit using the "extra port" usb cord mention in above update. Did not work. Did not even turn on. I think it will only turn on from USB power if only one drive is installed. Will have to carry power supply.

Legally purchasing music online.

2007-06-14T18:11:00.000-04:00

This online music scene has become a debocle with all the laws hyper-protective measures. Here I'll chronical the adventure.

Step one, learn to understand DRM. Not an easy feet. DRM is not really explained well on any music products sold nor by the online music vendors. DRM is an effort by the publishers to control distribution of their product. It will foil conventional attempts to copy the music in ways they deem illegal and will probably foil attempts by you to use it ways you think are allowed.

DRM comes in three main flavors. Flavor ONE; Subscription is "rented music". You don't own it. Usually you can play it on the provided DRM capable music player and most portable players. Stop paying subscription fees and you can no longer play the music. Flavor TWO; "Owned" music. I put owned in quotes because what you do with the music after you purchase it is still subject to the publisher's rules. Typically you can play it on the DRM capable music player and most portable music players. Flavor THREE; "Owned propietary" music. This is just like flavor two only the format of the music is propietary and can only be played on specific players and specific portable players. There are two major players in this area: iTunes and Zune.

Legend: S = Subscription DRM
O = Owned DRM
P = Propietary DRM

Service	DRM Types	Format	Sample Rates	Price	Notes
HMV	S,O	wma	?	£0.79	US$1.55 as of 6/14/2007
MSN Music	S,O	wma	256kbps	$0.99	KIA by Zune
Rhapsody	S	ra	?	$14.99/mo $12.99/mo
Walmart	O	wma	160kbps	$0.89
Yahoo! Music	S,O	wma	192kbps	$0.99	Bug infested

This is way too much work. FLUKE IT! Buy CD's and just rip the music from them. Good job DRM... you just killed your own market.

SSD (Solid State Drive) on a Budget

2007-06-01T15:13:00.005-04:00

USB Flash Drives

Can we use standard USB flash drives? We can essentially do this with normal hard drives through software RAID in the windows operating system. They must be configured as dynamic drives first. We see this option on our normal hard drives when we right click the volume on the disk manager. The option is not available to us when right clicking the USB flash drive volumes. So, we see what we should aim for in our first step! How to get the USB flash drive volumes treated as a normal volume?

Software RAID requires dynamic disk partition of drive. I think this only works on NTFS so the flash drives need to be formatted as such. I could not use windows interface to format the drives but the command line format worked with an odd behavior. The format would “hang” at the end. Pressing Cntrl-C would cause it to finish apparently without any errors. (?)

Did some searching and found some other people who where trying basically the same thing. One of the best examples I’ve read up to this point is from USBHacks (Forum Article).

There are a few things in my situation I don’t like about the listed “hack” above. One, I’m on Windows 2003 Server 64bit (and I’m attempting RAID0 for performance improvements so 32bit drivers do not sound appealing to me). Two, I don’t like the idea of using drivers meant for another device. It is interesting that Hitachi provided the c code to compile the driver mentioned in the hack!? (Not sure Hitachi actually wrote that or not?) But, the read was very interesting and does lead us in the right direction of where to look in the registry for comparing the devices as windows “sees” them. (Note: article used ControlSet001 and I use Current ControlSet… It’s my understanding that ControlSet00N are basically for the “last good known profile configuration” boot option? Not really sure about that so don’t take it as fact.)

Right off the bat, we see some differences in how windows 2003 is treating the keys in HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet Striped / RAID0 Two Western Digital Raptor SATA Drives running on Intel ICH8R southbridge. HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Enum\IDE\DiskRAID01.0.00__\4&amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;49adc61&0&0.0.0 The system drive is located under "IDE" (even though it is a SATA RAID0 configuration). Generic Microcenter 2GB USB2.0 Flash drive (only $12 as of 03/24/2007!). HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Enum\USBSTOR\Disk&Ven_USB_2.0&Prod_Flash_Disk&Rev_1100\AA04012700007752&0 The USB drives are located under "USBSTOR". Seagate 20GB Notebook drive in MaPower enclosure for USB portable drive. HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Enum\USBSTOR\Disk&Ven_ST94811A&Prod_&Rev_0811\001500013560&0

Change of Plans, CF Cards

Today read something that might bridge the gap to very cheap flash drive storage. Addonics makes a SATA CF reader. They also provide an interesting set of directions for configuring the compact flash as bootable. Apparently, this allows compact flash memory ( CF/II or CF type 2 storage compact flash cards according to the website ) to be connected via SATA ports. From there, they could be configured as drives and hopefully configured in a RAID configuration as well. Price shopping yields the Addonics adapter at approximately $35/ea. CF cards have highly variable costs. Speeds also appear to be highly variable. The fastest I could find in a brief search was Sandisk CF IV. They’re advertising 40 MB/s. Well, the SATA II / SATA 300 specification supposedly supports up to 300 MB/s. How does that stack up against a normal hard drive? I don’t expect the SATA hard drives to come anywhere close to that 300 MB/s bandwidth but I’m still curious how they actually perform. Quick jump over to a recent Tom’s Hardware article about hard drive performance shows something surprising. The hard drives, even at their WORST are performing nearly as well as the 40 MB/s. Err, WHAT!? I thought flash was supposed to be blazing speed! What about notebook drives? The hype about flash drives seems to derive mainly from their seek times. Seek time is basically how long it takes to find the point at which data I/O occurs. What are the seek times of flash versus current hard drives?

Okay, need to wrap my head around this. Bandwidth Ratings:

• USB = 1.5 MB/s

• USB 2.0 = 60 MB/s

• 1394a = 50 MB/s

• 1394b = 387.5 MB/s (3100 Mb/s)

• SATA I / 150 / 1.5 = 150 MB/s (1.5 Gb/s)

• SATA II / 300 / 3 = 300 MB/s (3 Gb/s)

• SCSI = 320 MB/s

• PCIe = 312.5 MB/s (2.5 Gb/s = 2500 Mb/s = 312.5 MB/s)

• PCIe 2.0 = 625 MB/s (5.0 Gb/s = 5000 Mb/s = 625 MB/s)

• Sandisk CF IV = 40MB/s sustained R/W

• Sandisk SSD 5000 = 67MB/s sustained R/W (on SATA 1.0a interface)

So, basically only first generation USB bandwidth would be a limiting factor for the advertised read/write performances we’re seeing from compact flash cards thus far (when one device attached). Now, what are the ACTUAL read/write performances reported from various devices. (Now that we know the communication standard is not limiting the device speed).

Read/Write Ratings: • Standard SATA II drives of recent manufacture are roughly 60 MB/s (read and write). • Generic USB 2.0 Flash drives are roughly 28 MB/s (read) 16 MB/s (write) at best and 1 MB/s (read and write) at worst. • Generic Compact Flash cards are roughly 40MB/s (read and write) at best and about 10 MB/s (read and write) at the lower end.

An excerpt from Tom's Hardware about memory sticks (article) : “Speed ratings are provided as a multiple of single speed CD performance, which is 150 KB/s, so an 80X flash device will deliver up to 12 MB/s, while a 150X device will do 22.5 MB/s.” This explains the “X” ratings seen on various vendor products (except for PCIe where it is a lane multiplier indicator like PCIe x16). The fastest I've seen as of this writing are 266X from Sandisk and A-Data and Transcend.

My thoughts are multiple CF 266X cards would provide the low seek time and high speed desired if put into an array configuration using the CF readers. I contacted Addonics and asked them to verify their reader has no bandwidth limitations inherit to it and they verified the bandwidth limitations are only based upon the media placed in the reader itself. Taking them for their word, that means I should be able to use approximately four of the Addonics readers with four CF 266X cards (SATA 1.0a yields 150MB/s. 150/40 = 3.75, or about 4 CF 266X cards). Technically, if all four cards where reading at top speed then the SATA I connection would be saturated but in a RAID 5 configuration, it will be slightly less due to parity operations. I do not expect the four cards to see any bottle neck in RAID 5.

Cards have been purchased. Readers have been purchased. I chose the Sandisk cards since they came with a lifetime warranty. Right off the bat there is a problem. The Addonics readers do not work with the ICH8R controller on the P965 chipset. Called the Addonics tech support and talked to a real person right away! They confirmed this non-functional problem. I asked if they had tested their readers on a SATA controller and they confirmed so on a SiI3112. I asked about RAID as well and they said yes but not reliably. I am disappointed this was not put on their website. Caveat emptor at its worst. You figure SATA 1.0a compliant would meant SATA 1.0a compliant regardless. Somebody isn't compliant I guess? (Either Intel or Addonics?) I did notice the reader would work if the SATA controller was put in IDE mode. I'm not about to cripple these cards doing that. It would appear the SiI3112 is only built into older motherboards? (Not 100% sure... not finding it in any new SATA controllers). I'm going to try a HighPoint RocketRAID 2300 PCI Express SATA II Controller Card ($109 from NewEgg). Hopefully, they're compliant and tolerant of lax compliance just in case.

Some reading about the removable bit on CF gives me some pause that these may never be "viewed" correctly by the OS. Apparently you have to pay much higher prices for "industrial grade" flash just to have the CF set as fixed disk / true-IDE. I noticed industrial grade also imparts higher temperature ranges and durability. NONE of those things concern me. Market deficiency is glaringly obvious for this particular need (cheap fast flash set to fixed disk).

Well the RAID card reads the CF cards just fine. BUT, all settings are wiped out upon every reboot. It would appear my fears about the removable bit are realized. Apparently, upon booting, windows looks at the drive and sees the removable flag set true. This means the drive may have been modified somewhere else so the system must re-initialize it. Thus, all array configurations are gone. With no array configuration, the data is gone as well. I need to find a way to set the bit right or else "trick" the OS into thinking the drives are fixed.

Discovered Transcend "Industrial" grade CF 266X (nearly 40MB/s). The "industrial" usually signifies removable bit can be set false. However, the cost is slightly more than their 44pin IDE drive of same size, 8GB ( $210 vs. $203 ). Both support up to PIO Mode 6 and Ultra DMA Mode 4 so they should yield similar speeds (all other bottlenecks remaining the same). I'm thinking this will probably be the route to take.

Looked into some specials at the transcend site and noticed fathers day special on the regular CF 266X mentioned the following: "Please Note: This card is programmed in DMA Mode by default. This card is capable of automatically switching between "Removable Mode" for CF applications and "Fixed Disk Mode" for IDE applications". Well, isn't that nice. Don't need the industrial spec any how after all (especially the silly temperature reqs). The item was discounted to boot ($116.99)!

The transcend cards arrived and I've hooked them to the Highpoint RAID controller. They retain their configurations between boots so I'm pretty happy about that. Read speed is nearly as fast as the SanDisk Extreme IV. However, their write speeds appeared to be pathetic! Write speed appeared to be probably 1/100th of the read. Such a lopsided performance seemed to indicate something was wrong. So, deleted everything and started over. Re-seated all connections, re-initialized devices, and re-created the array. I knew things were looking better when the raid initialization stated 3 minutes instead of 1 hour and it also formated in less than 30 seconds. So, now everything seems to be working good, its time to do some benchmarks.

	WD740 x1 (AHCI)	TS8GCF266 x4 (RAID5)	WD740 x2 (RAID0)	WD740 x4 (RAID5)	WD740 x4 (RAID5+Cache)	I-RAM
Burst	117.6	76.8	220.3	228.3	1969.5	134.6	MB/s
AVG	65	54.4	113.9	139.5	205.4	125.3	MB/s
RndAcc	7.9	0.4	7.9	7.4	7.4	0.1	ms
XPSP2 Boot	no attempt	19	15	na	na	na	s

Results strongly favor RAID0 configuration of hard drive despite the lower random access times of CF. Will post my thoughts on that when I get some time.

Things I'm reading related to this topic:

Is the CF interface natively more like IDE or SATA? (more IDE apparently)

Adapter for IDE to SATA and SATA to IDE

Adpater for IDE hard drive to SATA port (might use this with Addonics IDE CF Adapter? jeesh what a rat nest of wires that would be!)

And yet another one? (From same vendor?)

Perhaps its better to just skip the SATA conversion and interface with the CF cards at the more native IDE level? (less complexity = more stability?) This would require an IDE raid controller.

Hmmm... the "removable bit" might be problematic. (These guys also sell CF adapters for IDE and SATA). Sandisk number is updated to 866-270-5532. Tech support answered but didn't know anything other than that I should email OEMSales@sandisk.com . Emailing them to see if response prepared by monday. As of today (06/08/2007) Sandisk has not responded. I'm not holding my breath.

Read this article today (02/27/2008) about a new compact flash type named "CFast". It will apparently allow up to 375MB/s (it is not clear if that is limit from the newer SATA interface for the memory itself. I'm thinking the actual speeds due to flash memory limitations will be much lower. Article: http://www.dailytech.com/CFast+to+Replace+CompactFlash+in+Next+18+to+24+Months/article10842.htm

Comments are always welcome.