Incidents Per Device ?

[shaikh71]

Hi..

In my organization we manage servers remotely. We have a metric of measurement called incidents per device. I wanted to know what is the industry standard ratio is (My company has been assuming it to be 1.4).

Also if anyone could give me a breakdown technology wise like Wintel, Unix, Network, Novell, Exchange, Database.. It would be a great help.

Regards,
Shaikh

[Diarmid]

Is that 1.4 per minute?

If it is an industry standard, what happens if you drop below that figure?

Does it not matter what you do with the servers?

Do you expect a different figure if you have additional disc arrays in a server?

Would the breakdown by operating system be any good without separating the different releases and patch levels?

To take the extreme case (a thought experiment) if suddenly the figures appeared all nice and neatly summarised and tabulated for every server in the world for the last five years (including trend graphs and projections) what could you do with the information to help improve your service?

To make it truly useful we would have to extract only those systems doing exactly what yours are doing in exactly the same environment for exactly the same people in exactly the same way. But you already have that information (presumably).

By the way if it is per minute, then there is plenty of scope for improvement.
_________________
"Method goes far to prevent trouble in business: for it makes the task easy, hinders confusion, saves abundance of time, and instructs those that have business depending, both what to do and what to hope."
William Penn 1644-1718

[UKVIKING]

Actually

There are no standards for incidents per server nor is there for hardware either

You can ask the h/w company on how reliable their systems are - and they will produce specious data on coloured paper ({DAMN!! I have been borged into English - U's })

You can ask s/w company how reliable their s/w (o/s) are and you will get the same thing

Then you should look to your own kit - and do the statistics your self

Go by O/S, patch level, manufacture, system type as the classification
track the # of incidents, per day, week, month, and year

Dump the statistics data into excel and make several different types of graphs

Print out the graphs and spend hours analysising the graphs and present your finding to your boss
_________________
John Hardesty
ITSM Manager's Certificate (Red Badge)

Change Management is POWER & CONTROL. /....evil laughter

[Diarmid]

UK,

you are so much more positive than me I didn't think of the advantages of soaking up spare time in this exercise.

Don't tell the English (because they still remember Napoleon, and not fondly), but the u in colour is a French incursion into the language that took root a good while after the Mayflower set sail.
_________________
"Method goes far to prevent trouble in business: for it makes the task easy, hinders confusion, saves abundance of time, and instructs those that have business depending, both what to do and what to hope."
William Penn 1644-1718

[UKVIKING]

Diarmid

I made a NYR to be more nice and less ... well me on forums.....

well It is difficult giving the material I get to play with
_________________
John Hardesty
ITSM Manager's Certificate (Red Badge)

Change Management is POWER & CONTROL. /....evil laughter

[Diarmid]

John,

you mean sarcasm instead of rant (relative terms)? Next you will be resorting to irony!

The trouble is, once you start to get subtle, you risk being misunderstood, although that might just be a way of sorting the wheat from the chaff.
_________________
"Method goes far to prevent trouble in business: for it makes the task easy, hinders confusion, saves abundance of time, and instructs those that have business depending, both what to do and what to hope."
William Penn 1644-1718

[UKVIKING]

Diarmid

I risk being misunderstood anyway
_________________
John Hardesty
ITSM Manager's Certificate (Red Badge)

Change Management is POWER & CONTROL. /....evil laughter

[shaikh71]

Thanks for the info guys.

@Diarmid, The ratio of 1.4 is incidents per device per month.

Regards,
Shaikh

[UKVIKING]

is that a gregorian calendar month or a lunar month or a 30 day period ?
_________________
John Hardesty
ITSM Manager's Certificate (Red Badge)

Change Management is POWER & CONTROL. /....evil laughter

[Diarmid]

John, did you mean a rolling thirty day period? would that roll weekly, daily or hourly?

shaikh71, I was being just a tiny fraction pedantic about that. The fact that you failed to explain what your unit was seemed immaterial, because I cannot think of a single good use of a "standard" rate for incidents on a server.

1.4 (or any other number) is unlikely to be equally "right" for two servers performing different tasks and 1.4 per annum would be better than 1.4 per month.

The only relevant questions are:

- Is the customer getting the level of service they require and are paying for despite some incidents occurring? - if so, then the incident rate is "acceptable" in the sense that you have no need to panic. and so the second question kicks in:

- Can we do something to reduce the number of incidents in the service (continual service improvement, problem management and a few other things) and if so, will we get better returns from looking at the servers with the highest rates or will it be better to look at some other categorization of incidents?

Neither of these questions require some artificial norm for incidents per server. On the other hand tracking how many incidents occur on each of your servers can highlight anomalies that need investigating just the same as with any other CI.

Comparing to norms, whether real or made up, leads to complacency when your figures look good and panic otherwise. Whereas noticing that a server has had three times as many incidents this month compared to last month leads to investigation even if it is still under your norm.
_________________
"Method goes far to prevent trouble in business: for it makes the task easy, hinders confusion, saves abundance of time, and instructs those that have business depending, both what to do and what to hope."
William Penn 1644-1718

[UKVIKING]

In addition, the use of figures like 1.4 incident per server per month can then be used to provide useless babblespeak

the average number of incidents for the month of june doubled or halved

while this may look go from a bullet point of view (PPT), it is actualy meaningless

because you went from 1.4 average to .7 or 2.8 ... and the answer is so what

was the service impacted

You can have 1 incident that crashes the service for days - which skews the stats

look at things like floods, fires, earthquakes, terror attacks. the IT services in those cities had 1 big incident that was under their respective norm / average... yet the service became unavailable
_________________
John Hardesty
ITSM Manager's Certificate (Red Badge)

Change Management is POWER & CONTROL. /....evil laughter

[swansong]

Can I add that I have a Commodore 64 from Christmas 1983 and it has never once failed.

The incidents per device number must be something like a millionth of a billionth of an ounce, or even smaller.