Hi All,
I'm contemplating the detailed design of the 30-odd major sites that I will be
installing.
These range from hundreds of nodes down to tens of nodes in size. The biggest
will run on AIX or Solaris servers, but most will be on Windows. (As an aside
I'd prefer Linux, but it costs three or more times what windows does in support
costs). I'd like as standardized a design as possible to keep the complexity
down.
I've defined 5 "recovery levels" RL1 through RL5, and there will be five
domains available. A server with any data classified at recovery level 1 will
reside in domain RL1 and have access to mgmtclasses RL1 to RL5. A server not
in RL1 with any data classified at recovery level 2 will reside in domain RL2
and have access to mgmtclasses RL2 to RL5, and so on. Hopefully, only the
critical portions of the data on any given client will be classified at the
higher levels, and the rest will back up to lower levels.
Recovery level 1 implies a full server recovery (from a TSM point of view) for
a single server in 8 hours or less. This will require filespace level
collocation. RL2 is 24 hours and requires node level collocation. RL3 is 48
hours. RL4 is 72 hours and RL5 is > 1 week.
The idea is that the individual system admins can classify their own data. RL1
is for very critical production, 2 for Critical production, 3 for normal
production, 4 for test/dev and non-critical production and 5 for retired nodes.
Now the question is how to organize the collocation. I'd prefer from a wear
and tear point of view not to mount every tape every night (mostly SDLT), so I
was thinking of front-ending the collocated storage pools with non-collocated
pools, so that the collocation happens when the front-end pool gets to a
certain size, or maybe is scheduled over the weekend.
Now I realize that tape to tape pool movement is single threaded until 5.3
comes out, but other than that limitation, can anyone see any reason why this
would not work ? Will it have the desired effect? Is anyone doing this?
On another front, am I too optimistic to think that local client administrators
will appropriately classify their data?
Your experiences on this front too will be helpful - particularly those
universities out there who have little control over your clients.
TIA
Steve.
Steve Harris
Systems Admin Consultant
Enterprise Backup Project
Queensland Health, Brisbane Australia
***********************************************************************************
This email, including any attachments sent with it, is confidential and for the
sole use of the intended recipient(s). This confidentiality is not waived or
lost, if you receive it and you are not the intended recipient(s), or if it is
transmitted/received in error.
Any unauthorised use, alteration, disclosure, distribution or review of this
email is prohibited. It may be subject to a statutory duty of confidentiality
if it relates to health service matters.
If you are not the intended recipient(s), or if you have received this email in
error, you are asked to immediately notify the sender by telephone or by return
email. You should also delete this email and destroy any hard copies produced.
***********************************************************************************
|