Hi,
having started with ADSM in late 1997 and without any knowledge
of it:
> 1. How long did it take to get the ADSM server installed and
> operational?
It was already installed and ran somewhere in between test and
production when I got it handed over, although it was already
in production.
> 2. How long did it take to get the first 10 clients installed
> and operational?
See 1.
> 3. How long did it take to get 50% of the overall
> implementation complete?
To come to the point where we are today, it took me some months,
but as I also had to do miscellaneous other things, this should
have been not longer than 4-8 weeks all together.
The attendance of the Share conference in summer 1998 helped me
a lot in understanding ADSM better and review what I had done
so far.
> 4. How long did it take to get 100% of the overall
> implementation complete?
We're still not yet through - as we want to establish better
Networking capacities: Currently we run with Token Ring and
standard Ethernet and will change to fast ethernet by late 1999.
Some additional issues regarding tape technology and assigned
capacities are also ongoing.
> 5. Were you and your management happy with the overall
> implementation time?
No.
> 6. What would you have done differently?
I see a need to do it this way:
Step 1 - Planning:
a) Get figures of what You want to backup/archive first.
Estimate high and add 10-25%.
b) Then set the expiration dates for archives and for the
backups the required number of generations.
As this is one of the key issues, You will have to go
and ask the end-users and also plan with them for
service level agreements.
Get and idea of what backup/archive may run at what
time.
c) Look if the network can cope with at least 100% of what
Your estimates are today and plan for 150-200% of this
initial value.
d) Classify the ADSM clients (critical, important, less
important) and plan for the necessary ADSM server
availability, including a Disaster Recovery of the
ADSM server.
e) Get an idea of how much outages or loss of data will
cost.
f) Plan the ADSM server(s), their platforms, locations,
the tape technology and capacities.
g) Calculate the cost and see whether it is balanced. This
is also the time when to fix the planned service level
agreements from 1b), as now You know what the whole will
cost You and whether it IS balanced by the data from 1e).
Step 2 - Preparation:
a) Get all involved people educated. This is one of the
key issues to success with ADSM as You not only have to
read and understand the reference manual - You also have
to adopt the philosophy of ADSM.
b) Set up a test scenario with all possible client
platforms now and plan for the final production
environment.
c) Assign responsibilities and backup persons.
Also a key issue, as this will have to be known and
lived later on (as ADSM tends to become a forgotten
child somewhere in the corner once it is up and
running).
d) Plan for operation, monitoring an reporting; look at
possible tools from 3rd parties for that.
e) Assign the locations for the servers, tape environment
and offsite storage.
In case You plan for redundant ADSM servers and tape
libraries, separate these as much as possible to avoid
disaster impact to one location.
f) Establish a Disaster Recovery Plan for the servers and
the clients.
Step 3 - Implementation:
a) Do tests with the test scenario from 2b).
This includes monitoring of capacities and duration of
backup/archive/restore/recall.
Include a bare metal restore of the largest ADSM Client
to know how long it will take at least.
Review that and see if it can fulfill the service level
agreements - take apropriate (cost balanced) action.
b) Implement the production environment and keep the test
environment for later tests (Y2K, fixes, upgrades, etc.).
c) Document what has been done in a ADSM server maintenance
log, as ADSM does not keep that forever if at all.
Key issue is here, that You will forget WHY things have
been done this way or that way and You will go through
possibly painful experiences again.
Step 4 - Production:
a) Put everything in place and:
- Review resource usage
- Monitor capacities
- Do daily reports and after 2-3 months when the load
gets balanced, try to do forecasts (for ex. we have a
observed a steady growth of client data in the range
of 50-120%). So You can plan ahead and act in time.
b) Review the
- duration of client sessions
- fullfillment of the service level agreements
(this includes esp. disaster recovery tests and full
restore tests)
- cost
- disaster Recovery procedures and test at least once a
year.
In the end it will make the environment reliable, cost
balanced and You have satisfied end users as they have
taken the decision of what they want to have and what that
service level may cost.
> 7. If the overall time to implement was overly long, in
> your opinion, did this time cost your company too much
> money?
Yes.
> 8. Is there value in having a shorter ADSM implementation?
It's up to You:
Either take time for planning and prepare carefully - and
You will get an excellent return on invest.
Or put it in as a quick shot and go through all the bumps
and painful experiences this environment has for You if
You don't take time to plan ahead.
Best regards
Andreas Peikert, Systems Programmer
aMaDEUS Data Processing GmbH
P.O. Box, D-85424 Erding, Germany
Fon: +49 (0)8122 43-5805
Fax: +49 (0)8122 43-3260
email: apeikert AT amadeus DOT net
peikifex AT peikert DOT de
<* save a megabyte - do you really have to file this note *>
|