System Overview

[Sharon Bi]

Dear all,

Just come to a new environment. There are 21 production boxes. Mostly K class, one two-node N class cluster,EMC, with HP OpenView ITO in place, using ADSM to backup....

We were asked to provide a system overview report and come to some recommendations to improve.

I have a list of the following information: network layout, file system structure, peripherals, lan connections, cron jobs, disk layout, database... Is there anything else I should put into the list? And which area should I look into to make improvement?

I know its a big question... but I just don't know where to start...

Thank you very mcuh!

Sharon

[Berlene Herren · Accepted solution]

How about versions of software, firmware level and patch level?

Just my two cents.

Berlene

[MARTINACHE]

Hi,

I think that adding some simple stats (load, sar, vmstat, users cnx, ...) is also interesting.

With this king of data, you can evaluate system use.

Regards,

Patrice.

[Dan Hetzel]

Hi Sharon,

Ralph Roth posted the new version of a nice tool called 'confi2html' which will surely help you in creating nicely formatted HTML reports for all your systems.

Here is the link:

http://forums.itrc.hp.com/cm/QuestionAnswer/1,1150,0x846637f45ef7d4118fef0090279cd0f9,00.html

Best regards,

Dan

[James R. Ferguson]

Hi Sharon:

Things that come to my mind, immediately, include:

> Installed software (online JFS, MirrorDisk, Glance, etc.)
> OS patch levels (General Release bundle)
> Diagnostis installed & level (STM, Predictive Support, EMS)
> Consistency (or not) of uid/gid across platforms
> DNS, NFS, NIS
> Backup strategy solution

...JRF...

[Devbinder Singh Marway]

Devise a list what are the most common problems you have encountered with the current setup. From this list you will see a lot of improvements or unless you have systems with no problems what so ever.
On the network side , see how secure your connections are i.e. do you have firewalls in place , which block user connections or can you get direct to the servers, do you have user authenticity . What monitoring tools/ scripts have you got to monitor the network i.e. what happens say if a device on the network cannot be seen, do you get alerts.

disk layouts , if you have databases in place what is the service level agreement 24x7 , how much downtime do you have . I read you have ADSM installed , this is a big area in itself , i.e. what monitoring tools do you have in place i.e. check failed schedules, check DB utilisation, check for scratch tapes in use, Have you done a DR on adsm , how long will it take to restore the data ?
etc..

Cron jobs which jobs take a long time to complete and use a lot of resources , if possible run these at a later time.

The other area is change management , do you control what is sone on each server , or can system admins do tasks on their as they please.

what about High Availablity , do you have it in place and how resilient are your systems. How often do you install patches recommemded by HP , do you have predict installed ? Hp will monitor h/w for you and notify you of problems or disk firmware upgrades.

Hope this gives you a bit more info , the main improvement is monitoring the systems and prompting you before the users realise that there is a problem i.e. simplest one is space , have thresholds set , and if the threshold is reached send a alert so you can act upon before there is a real problem.

[Stefan Schulz]

Hmm... i would think if you are asked for recommendations you should know where they think the bottlenecks are.

So i would build a list with all the services, databases and so on running on those boxes. then i would look for poor performance on this services.

Ask if the users have to do routine work by hand, perhaps you can optimize by better systemlayout.

Are there any applications which should share data but don't? Perhaps they can share data with a better filesystem layout.

Hope my english is good enough.

Regards Stefan