chris harris [msft]

I'm the PM that owns the Exchange Management Pack at Microsoft. I'm working
on the reports for our next major release and I need your help.

I want to make it easier for you to report on service/server availability
and to measure your compliance with Service Level Agreements.

Please tell me how your orgainization measures availability or defines
Service Level Agreements (SLA) for Exchange. Or, perhaps more importantly,
how you would like to measure or define them.

Examples:
* Outlook clients must be able to successfully connect to the server 99.5%
of the time during normal business hours.
* The Exchange databases must be mounted 99.9% of the time


Please provide the following:
1. Current SLA and availability definitions
2. How do you measure it today?
3. Preferred SLA and availability definitions
4. How would you like to measure it?

To make it worth your effort, I've got a number of different software
packages sitting in my office, and for the best idea I'll let you take your
pick.

If you

Thanks,
Chris Harris

blake mengotto

5 nines!! Just kidding, I'll send this to management, so they can never
read it and then ask me why it wasn't a built in feature of MOM and the
Exchange Management Pack two years from now. Then I'll tell them to check
the message I sent, but they will blame mailbox manager for grooming it out,
and ask me to restore their mailbox from tape only to find out the server
farm inserted the tape upside down in the library for the past two years, so
there is no good backup, then ultimately it's my fault for not "owning" the
project, resulting in me getting fired and then trolling these message
boards for the remainder of my years as I get certified in Quantum Fri
Mechanics for my big interview at McDonalds.

<3 Blake - The MOM Misfit

teo gomezµê¶)ïz{k‹h¶w­

I would love to be able to run a report that tells me what servers have had a
full backup done and which ones haven't.

--
Teo Gomez

chris harris [msft]

I like milkshakes with my fries - especially Egg Nog ones at Christmas time.
Can you make me one of those as well?
--
Chris

lars norman søndergaard

Chris,

First of all: thanks for giving us a chance to influence the next version of
the Exchange Management Pack. In terms of SLA here is some of the things we
would love to see:

External SLA (that is: numbers that our customers see)

1. Internet mail availability. Basically like the mail flow sender/receiver
scripts but going against an external server that will simply bounce the
mail. Currently we have done this using a number of home-made scripts and a
server placed on a ADSL line. I all honestly we are really measuring the
ADSL availability. Perhaps Microsoft could host a few "reply servers". Since
external mail is beeing scanned for contents and AV we would probably love
to have different thresholds for day/night.

2. Availability for various clients: ability to send mails using Outlook and
OWA. It would be very beneficial if the availability could be broken into
office hours / non office hours. Since we are a world wide operation this
should take the local timezone into account.

Internal SLA (that is: numbers we measure our selfes against)
1. Backup successfully completed within the backup window
2. Store availability
3. Services availability
Internal SLA are measured on a 24 hour schedule.
Internal SLA is probably going to be replaced with the "MSIM SLA Scorecard"
when released.

In terms of other issues with the current management pack:
1. The sender/receiver scripts do not take into account that some locations
have "strange" timezones (Kathmando GMT+5.45). This is not really an issue
with the MP but rather that MOM is missing a GMT timed event provider.

2. Performance of the sender script is not very good. We have a few
connector servers from where the mails to each of our branch offices are
sent. We see around 10 mails sent / min.

Reporting
Reports for SCRM or the DW showing
the values for the last month availabilty/day
the aggregated value compared with the last 12 months.
A graph showing how much better we are doing this month compared to the last
(perhaps a script that will make our historical availabilty worse so our
current data look good ).

Thanks
Lars Norman Søndergaard
lns@globeteam.com
+45 23707679

"Chris Harris [MSFT]" <charris@online.microsoft.com> skrev i en meddelelse
news:evENjr8ZFHA.2496@TK2MSFTNGP14.phx.gbl...

chris harris [msft]

Lars,

This is exactly the kind of information I was looking for. Thank you for
taking the time to write it up for me. You are certainly in the lead for
getting your choice of the software. I'll let the "contest" run through the
weekend and make my decision of who is the winner next Monday.

A couple of questions regarding your items below:

*External SLA*
General - Do you anticipate the need to automatically (or manually) adjust
the numbers to account for maintenance windows or to provide an explanation
of what caused an outage that will appear in the report?

#1 (Internet mail availability) Since we don't know what "day/night" means
for each organization and I feel I can safely assume it's not 08:00-17:00
and 17:00-08:00 would it be acceptable if we shipped one rule that covered
24x7 and you could override and/or copy it to set up thresholds for your
companies business schedule?

#2 (Client availability) Do you have any offices that operate with multiple
shifts? If so, how would you like to see the day broken up to account for
the shifts? Or would you prefer just to show the availability for 24x7?
Do you view client availability as the ability to send a mail or as the
ability to simply connect to the server?
Do you have any requirements for client performance? (i.e. 95% of client
transactions must complete successfully within 30 seconds)
Is it important to verify that an actual Outlook client can logon or would
it be acceptable if the test simply determined if a MAPI logon can be
performed?

*Internal SLA*
#1 (Backup) Could we safely assume that all backups should be completed by
06:00 local server time?
#2 (Store availability) What does store availability mean to you? Does it
mean that just mean that the Information Store service is running? That the
database are mounted? That if a client were to attempt a logon that they
would be able to do so successfully and read their email?
#3 (Services availability) What does services availability mean to you? Just
that critical/required services are running?

Again, thank you for the feedback,

--
Chris Harris
Microsoft Exchange

****** Disclaimer ******
This posting is provided "AS IS" with no warranties, and confers no rights.
Use of included script samples are subject to the terms specified at
http://www.microsoft.com/info/cpyright.htm.
****** /Disclaimer ******

lars norman søndergaard

Hi Chris,

Again I am really glad to able to get some of our ideas considered for the
next version. It is so much easier than creating the rules your self.

To answer your questions:

1. External SLA/general: We are probably going to do a report aggregating
the data from your new rules (!) together with line availabilty and
availabilty of the file server. We will manually (unfortunately) create a
cover letter describing major incidents. Since the current version of MOM
does not allow to grab a number of alert and assign them to an incident we
can not simply query the database. One idea we have been kicking around was
to use a special tag in the description field when we resolve alerts and
then include these in the SLA reports.

We are considering a layout that looks a little like this (I can probably do
a .RDL mockup if needed):

Office in Hong Kong:

Server Availability
This month
Date: 1 2 3 4 5
Office hours: 99.9% 99.7% ...
Non-office 100% 98% ...

And the same for Exchange Client Availability & Internet Availability and
for file share avail.

Then show an aggreate
this month last month month-2 ....
month -12
Server Avail (office) 99.902% 99.821% ...

Perhaps a graph here

Some boilerplate text describing how the numbers are measured and how they
should be interpreted.

2. External SLA/1: You are right, it should be done wia overrides. It would
be great however if the time dimension from the SCRM dataware house could be
used so we do not have to specify the business hours for each branch office
both in the rules (we have three MGs) and in the DW.

3. External SLA/2. No we do not have multiple workshifts. But we would like
to break up the SLA in business/non-business hours since we are really
focusing on providing services 07.00-18.00.
We do see client availability as the the actual ability of an Outlook client
to send/receive mails. But i think we can assume that if a MAPI logon is
possible then Outlook should work too. The flow of mails as soon as the mail
hits the local Exchange server is not important - that is covered by the the
mail flow scripts (and it is almost impossible to provide any good SLAs on
this - we have 85 sites and if we for a given site was to report availabilty
to send a mail to the other 84 site that is just too hard - and makes little
sense).
In terms of reporting on client performance: it is something to consider for
later (for us). I am sure that users would be interested but I would rather
wait.

Internal SLA
1. Our backup window is from 22.00 to 06.00 (local time)
2. A combination of the store service running and it beeing in a healthy
state. Really I think we could say that if the store service is running then
the store is fine...
3. Exactly! that services are running.

As I said we are probably going to structure of internal reporting on the
MSIM SLA Scorecard so rules/reports provided here will probably be used. The
whole idea is that the internal SLA should be used to plan hardware
upgrades, find new processes and so on - and hopefully make the external
SLAs look better.

I am probably going to create a PSS case for some issues we have with the
current Exchange MP. If you are interested I can send you the case number?

Thanks
Lars Norman Søndergaard

"Chris Harris [MSFT]" <charris@online.microsoft.com> skrev i en meddelelse
news:%23f6wRk3aFHA.2996@TK2MSFTNGP10.phx.gbl...

chris harris [msft]

Drum roll please...

And the winner is "Lars Norman Søndergaard"!

Out of the millions of entries submitted, Lars' was chosen by me in a
completely non-scientific way as the most helpful for our plans in the next
release. I have contacted Lars offline to let him know that he won and to
provide him with the list of software packages that he can choose from.

Thanks to all who participated!

Chris
--
Chris Harris
Microsoft Exchange

****** Disclaimer ******
This posting is provided "AS IS" with no warranties, and confers no rights.
Use of included script samples are subject to the terms specified at
http://www.microsoft.com/info/cpyright.htm.
****** /Disclaimer ******

tony gn‰òجrë,Š‰ìš+¢ÊµÊ&

Chris,

I have what may seem like a simple request to you but, I am not a developer.
I am an MS Partner and ALL of my clients want the Exchange availability
report but they want to be able to leave out a predefined maintenance window
since it is outside of their SLA's. Regardless of what their SLA is, they
all want to exclude the maintenance window.

Many of them also would like to be able to choose the granularity of the
report (per store, per server, per management group, per og).

Regards,
Tony G

chris harris [msft]

Tony,
Excluding maintenance windows is something we are considering. However, I
would be very interested in hearing how your customers feel it should be
implemented? Do they want to somehow manually specify the window? Does the
window change? Do they want to just exclude the time that the MOM agent is
in Maintenance Mode? Etc.

Thanks,

--
Chris Harris
Microsoft Exchange

****** Disclaimer ******
This posting is provided "AS IS" with no warranties, and confers no rights.
Use of included script samples are subject to the terms specified at
http://www.microsoft.com/info/cpyright.htm.
****** /Disclaimer ******