Windows 7 Reliability Monitor

Let’s trade one Reliability Monitor mystery for another, OK?

Anybody who’s been reading these blogs for any length of time knows that I am enamored of the Reliability Monitor in Windows Vista and 7. Back in early September I wrote a blog entitled “Why doesn’t Windows 7 post a reliability index any more?” At the time I was (and remain) a bit miffed because MS has to store a stability index value to graph out the basic reliability history, yet it chooses not to explicitly display that value when it draws out the graph for your information and edification. Here’s a somewhat squeezed down display from my system this morning, so I can point out a few bits and pieces.

Today, the stability index is less than great, but better than awful

Notice the absence of numerical values for the stability index anywhere on the display, and the categories for errors (red X), warnings (yellow exclamation point), and information (white “i” on a blue background) that the utility reports day by day. Note also the links at the bottom of the window.

As it happens, I jumped into a great blog by the inimitable Ed Bott at ZDNet this morning entitled “Windows 7 in the real world: 10 PCs under the microscope.” It’s a fascinating recitation of his experieces with 10 different PCs running Windows 7 over the past year or so. Throughout he cites specific stability index values from those machines, and I found myself asking “Where is he getting this data?”

That’s when I returned to the tool, and clicked the link at the lower left of its window (in case you can’t make out what it says, the link reads “Save reliability history…” I clicked the link, and saved the resulting data as an XML file, which I then opened in Internet Explorer. Bingo! As the following snippet illustrates, this is where you can find the actual numerical stability index value for any given day, captured at hourly intervals.

The graph data is right there in readable form

If you really want this data, you can go get it for yourself. But my question is: given that MS has to read this data to draw the graph in the first place, why can’t they add the few necessary lines of code to report the value in readable numeric form. They did it for the Vista version, but not for the Win7 version. I’m confused… and I hope they decide to fix this in an update or Service Pack some time soon.

That said, Bott also makes some great points about the value and meaning of the stability index in the aforementioned blog, on page 2. I’ll quote it verbatim:

And yet… My experience with this machine has been overwhelmingly positive. It runs nearly everything I throw at it and has no annoying bad habits. It doesn’t crash. It sleeps and wakes up reliably. The Reliability Monitor algorithm deducted huge amounts from the stability index (a total of more than 6 points) for two incidents that consumed 2-3 minutes each. In once case, an IE8 tab crashed four times in the space of a minute or two because of a problem with Adobe Flash in a single tab. Solution: Close that page. Two days later, I updated the excellent MediaMonkey music organizer/player to the most recent release, which proceeded to crash when I tried to run it. A quick trip to the support forums turned up the cause (an incompatibility with an iTunes 9 component) and the temporary fix (renaming a DLL). Although each event was annoying for a few minutes, neither one had even the slightest impact on performance after it had passed.

Lesson learned? If you’re happy with the way your system works, don’t obsess over a perfect 10.

On my production machine, I’ve had similar problems with several bits and pieces that have cost my stability index dearly. Until Dell came up with a bona fide Windows 7 driver for my AOL 968 combo device (print/scan/fax) I might see anywhere from three to ten “Printer Filter Pipeline Host stopped working” errors a day. Likewise, PC Doctor (the new beta version that supposedly works well with Win7) includes PC Tools Security Service item that crashed 7 times on 10/7/09, leading to the precipitous drop you’ll see in my stability index earlier in this blog. I’m happy to report that a switch to NIS 2010 took care of that problem, and that everything’s quiet with the AIO 968 since I upgraded the driver on 10/14. And now, I guess, I’m going to learn to live with a less-than-perfect stability index as long as my problems aren’t too serious or vexing.

Hmmm…This is an interesting Win7 situation

Those who’ve been reading this blog for any length of time know I’ve been fighting with a balky, or perhaps blighted, production PC for some time now. I upgraded this machine to Windows 7 in early August in hopes that it might help with chronic instability issues. Since then, results had been mixed with reliability index values fluctuating between 4.8 and 7.0 or thereabouts (in Windows 7 it’s kind of hard to tell because the Reliability Monitor no longer reports numeric values for that index, so you must eyeball it from the scale at the left of the graph). But over the past 10 days my system has experienced what mathematicians call a “montonically increasing” pattern, which translates into plain English as “upward and onward.”

Reliability Monitor as of 9/4/2009

I can’t tell if I’m at a 10.0 value, or simply approaching that value, but this is probably the highest reliability index I’ve ever seen on this machine. And of course, that raises a very interesting question: should I just leave this machine alone and keep on going with what I’ve got, or should I tear it down and install the processor and other components on the new Asus P5Q3 motherboard I ordered a couple of weeks ago to replace the Gigabyte P35T-DQ6 motherboard that’s currently in there (I had also planned to move the system into a bigger, better-ventilated Antec 900 case, and use an SSD for the boot drive)? I’m mindful of the old saying “If it ain’t broke, don’t fix it” — we are talking about my production machine after all. But then, I start to recall all the aggravation and heartbreak this machine has caused me in the past year, and wonder if this isn’t just a lull between storms.

I’m probably going to sit tight for a while because I’ve got to build a machine for my wife, for whom I just bought a barebones compact system built around an MSI Core 2 Duo Mobile motherboard (945GME1 mini-ITX model) and a Morex 150-Watt compact mini-ITX case with an Intel Core Duo T2300 processor (1.66 GHz). I’ve got SO-DIMMs out the ying-yang, and a 5,400 RPM SATA 2.5″ 160GB drive scavenged from my MSI notebook ready to install in that unit. I think I’ll do that system first, and also install my new Samsung monochrome laser printer, and use that interval to see if the good behavior from my production system is a glitch or the real thing. And if it ain’t broke, I may very well NOT fix it, but may take advantage of sub-$200 prices on Core 2 Quad Q8XXX CPUs to build a new test system re-using the case and power supply from my wife’s now-retired Sempron 3200+ system that gave up the ghost a couple of weeks ago.

More on this as the situation develops. Tell you what, though: I’ll be astonished if my production system has finally settled down. We’ll see…

Why doesn’t Windows 7 post a reliability index any more?

No numeric index value appears in the Win7 Reliability Monitor

Take a look at the Reliability Monitor output for my production PC, which has been running Windows 7 since my birthday on August 8. In addition to showing why I’ve got a new motherboard waiting for me to have time to swap out with its current Gigabyte P35T-DQ6 motherboard, it also shows no numerical values for the reliability index on any given day. I’ve highlighted the most recent Windows OS failure on 8/19/2009 by way of illustration, but what I want you to note is that while you can eyeball an approximate value for the RI (reliability index) on any given day — 8/19 looks like about a 6 to me, for example — the utility no longer reports a number anywhere in its display. Of course, that number has to be around somewhere, because the utility couldn’t draw the graph without numerical data from which to plot the points.

On my Dell D620 notebook running Vista Business, however, Reliability Monitor is happy to show me an index value for any day I select, as indicated here:

Notice the Index number value at upper right

During the beta period I tried like the dickens to contact the MS beta team and tell them to please add this back in, but I’ll be darned if I could ever figure out how to get into the proper channel to deliver that feedback, let alone share it with somebody who could actually *do* something about it. Now, all I can hope is that somebody at MS will see this lament and get it scheduled for Windows Update inclusion. I think we all deserve to know what our RI value is, rather than having to guess! What say you?