[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [SAGE] Strategies for taking ownership of existing infrastructure?




 Hi Jesus,

 when starting my most chaotic job as a sysadmin there where just some 
aspects
of the challenge you have to cope with... but I think some of those will 
be quite
handy:

 - set up monitoring (if not in place.. which I would assume)

    It doesn't sound like a quick win in first place... but setting up 
some sort of monitoring covering
    all your machines will help you get a first overview of

       * what you have
       * what might be the difference between "similar machines"
       * which are the big troublemakers (even if your customer did not 
yet recognise them to be)

    For a start I used the free version of big brother, which comes with 
valuable agents
    out of the box... and (for solaris) even with reasonable 
threshholds. The best thing for
    me was the grouping of machines belongin together logically.

    The monitoring can represent a first kind of "asset/configuration 
management"

    The next thing I tried was having all alarms directed to my 
cell-phone... ginving me a REALLY
    good overview of "what's going on" and "which are the permanent 
trouble-makers"... but lead
    to about one year of "just work no fun".. ;-)


 - set up install servers (if not in place.. which I would assume)

    An already chaotic environment normally tends to quick evolution... 
you should be able to cope
    with new machines and services in a manner that you control... don't 
be the disk-jockey...
    do things once on your install server

For further ideas it would be useful to know in what kind of 
organisation you are working...
how many people and how many machines you are about to take over... (I'd 
assume you don't
have to be SOX-compliant ;-))

 Good luck!

 Thomas

 



   

Steve Barnet wrote:

> Jesús Couto wrote:
>
>> Hi.
>>
>> The subject line sound awful (and its my first post :-/). Basically
>> I'm just asking for advice for the following situation:
>
>
> Welcome to the jungle. :-)
>
> [...]
>
>> I guess people here have found themselves in the same situation... I'm
>> trying to organize my ideas at all levels (technical, practices &
>> procedures, organization, "office politics") about how to get to own
>> the place. So any tip or advice you can come up with is welcome.
>
>
> In the past I've found it helpful to do a couple things:
>
> 1) Start work on a simple network map. It should cover physical
>    and logical components. You might need a couple of documents
>    to cleanly represent the information.
>
> 2) Make a plan. Make your "this is messed up" list, and then
>    make a "how it should be" list. The plan is the how you get
>    yourself from "messed up" to "how it should be." It doesn't
>    have to be fancy, but it should be a list of projects broken
>    down into steps you can cross off as you go along. Doing this
>    helps keep you on course, and gives you a way to keep track
>    of what you've accomplished.
>
> You can organize the plan in whatever way makes sense to you.
> I've found the following to work well for me (in rough order of
> system impact):
>
> Physical
>   Find and label everything - if you can't find it, you can't fix it.
>   Environment issues (power, cooling, noise, physical access, etc)
>   Cabling - bad cables make life miserable
>
> Network
>   Switches/hubs/routers etc
>   DNS
>   IP address management
>
> Application/Service
>   Email, web, etc (organization dependent)
>
> Procedural (How you and the organization work)
>   Problem tracking/reporting
>   Change management
>   Administrative access
>   Internal busimess processes, etc
>
> I've found that problems tend to percolate from physical ->
> network -> application/service in some rather subtle and nasty ways.
>
> A few other thoughts in no particular order:
>
> * Identify critical services (things that stop business, things
>   that get you fired). Protect them as though your job depends upon
>   it. :-)
>
> * Tread lightly at first. Take time to verify that what you're about
>   to do won't kill something else.
>
> * Celebrate small improvements. In many cases, improvement happens
>   as a result of small steps. When you're overwhelmed it's easy to
>   overlook the cumulative effects of small fixes.
>
> * Talk to your users and managers. It's quick to conduct business via
>   email, but many times you can find out more in a five minute
>   conversation than in a 20 message email exchange.
>
> * There comes a point in any migration process after which you need
>   to actively hunt down the stragglers and migrate them forcibly.
>
> Of course, the list goes on. Hopefully this can get you started.
>
> Good luck!
>
> Best,
>
> ---Steve
>
>