They say with hard disks, it is not if
your disk is going to fail, but rather when your disk will fail.
Anyone who has gone through a disk failure knows just how painful the process
Of course, the most important line of defence
in ensuring a disk failure is as painless as possible is to have a good
backup strategy. For maximum ease and convenience, a complete backup of
all hard disks on your computer is ideal. As a minimum, all documents should
be safely backed up.
But, wouldn’t it be nice to know your hard
disk was about to fail before it died an untimely death? Enter DiskAlert,
a new program from Executive Software. It is designed to constantly monitor
your hard disks for errors. If you start getting errors, DiskAlert can
notify you. The idea is that you can replace a failing hard drive before
complete and catastrophic failure. In addition to monitoring for disk errors,
DiskAlert can monitor the amount of free disk space.
DiskAlert can monitor just about any type
of hard disk—IDE, SCSI, and both software and hardware RAID arrays. It
only works on Windows NT, 2000, and XP, but you know my feelings on operating
systems. I have not run Windows 9x (on any system I cared about) for almost
2 years now.
DiskAlert has two thresholds you can set—warning
and critical. When DiskAlert reaches one of those thresholds, it notifies
you. It can do this with a pop-up window, e-mail, pager, or even a phone
call. For phone calls, you can have it play a WAV file. Nifty! You can
have completely different notifications for warning thresholds than those
for critical. There is a global configuration for all drives. An over-ride
can be set on any disk to have different thresholds or notifications. You
could set an operating system disk to notify at a very low threshold of
errors and a disk with games installed to a very high threshold. Or vice-versa,
depending on which you value more!
One of the nice features about DiskAlert
is that you can configure remote agents on any or all the computers on
your network (licenses are required for each disk) and you can have a single
console configure all machines.
When you call up the program, it loads
as a Microsoft Management Console snap-in. The left pane shows the workgroups/domains
being monitored and below each of those, the computers being monitored.
The right pane is split horizontally, with the top having a listing of
the drives being monitored and the bottom showing the status of the selected
drive. It is this last pane that proved to be the biggest pain for me.
You can switch between showing Disk
Health and Disk Space Used. When you choose Disk Space
Used, a pie chart shows the current usage. This is no more than
Windows Explorer will show you, other than the fact that the chart will
be green is if usage is below the warning threshold, yellow if it is between
the warning and critical thresholds, and red if it is above the critical
Disk Health shows as a line graph,
and is confusing. There are yellow and red horizontal bars representing
the warning and critical thresholds. A thin red graphing line shows the
error rate, and a thin blue line shows the disk throughput. I never saw
any errors on my drive, so I can’t comment on how the graphing of errors
works. But for disk throughput, there appears to be problems with choosing
ranges to show. While you can select ranges of all, today, this week, or
this month, no matter what I choose, the graph doesn’t change at all. It
appears to always show the disk throughput for just the current day.
But the problems with the graph do not
stop there. I was surprised to see the blue graphing line so low on the
chart. Then I noticed a tool-tip type window appear when I hovered over
areas of the graph. It detailed the exact throughput at a particular point
in time. It turns out that the graph is actually upside down. Any spikes
in the graph represent a decrease in throughput. The help file does
explain this, but it does not explain what the numbers in the scale to
the left of the chart mean. On my system it appeared to be a logarithmic
scale with 0 at the bottom, 6.84 about 1/3 of the way up the scale, 46.78
about 2/3 of the way up and 320 at the top.
When I first tested the email notification,
it wasn’t working. There were no error messages. Clicking the help question
mark in the dialog box used to define notifications and clicking on various
parts of the dialog box did not bring up any help text. The help file,
when opened manually did not tell me anything. I discovered that the Sender
needed to be in an SMTP format, such as DiskAlert@opcug.ca, rather than
something like Drive-C. I did not get any error message when I tried
to use Drive-C and then clicked on the Test button. It just
didn’t work. Note that it doesn’t have to be a valid SMTP address. It just
has to look like one. Once I configured the Sender to look like
an SMTP address, email notifications worked fine.
When I first started testing DiskAlert,
I found that my hard drive would grind away like crazy every minute or
so. I was beginning to suspect Disk Alert would cause a premature
death of my hard drive rather than just left me know it was about to happen!
When I got an updated build (188.8.131.52), this problem went away.
DiskAlert requires Windows NT (SP5 or greater),
Windows 2000, or Windows XP and Internet Explorer 5.5 or higher. It uses
only a couple of megabytes of disk space. On my system, which has 512MB
RAM, the three services (DiskAlert Agent, DiskAlert Licensing, and DiskAlert
Notification) took 14MB of RAM.
I struggled to find a reason to purchase
DiskAlert. I added up the pluses and came up with the following: DiskAlert
will notify you when disk space gets low or if your disk starts getting
errors that are higher than a certain threshold so you can replace the
disk before total failure. You can have a single console view to your entire
organization of multiple computers and hard disks.
On the minus side of the equation, I have
the graphing that does not seem to be working correctly. On critical systems,
most people run RAID 5 arrays that will allow complete failure of a single
disk without loss of data. Given that DiskAlert starts at US$50 for a single
license and drops to US$15 if you buy more than 1,000 licenses, it seems
out of proportion with the cost of disks these days. I compare it somewhat
to the cost of extended warrantees on consumer electronics (which I think
also tend to be way over-priced.) If you think of them in the 10% to 15%
range, that would translate into DiskAlert being priced at about CAN$20
to CAN$30, not CAN$75. I know that this sort of comparison is not really
comparing apples to apples, but still. Also, consider that DiskAlert is
licensed on a per disk basis, not per computer!
DiskAlert is probably really targeted more
at the enterprise than individuals. But even there, I just don’t see it
as being a cost-effective solution. Most enterprises run RAID5 arrays for
servers. The loss of a single disk is not a problem. Really critical servers
have hot standby disks. In the event of a failure of one disk in the array,
a rebuild begins immediately on the standby disk, reducing the likelihood
of a second disk failure causing downtime.
Having said all that, I suspect DiskAlert
is worthwhile for the truly paranoid or for those who can not afford to
take any chances at all. If you absolutely must know about an impending
disk failure before it happens, it is probably an effective tool.
You can download a 30-day trial version
US$50 for single license
to US$15 for more than 1000 licenses
Originally published: May, 2002