On 2/20/2014 6:23 AM, Jack Kingsley wrote:
I have a small group of files that I need to start to journal. Is there a
way to determine how journaling of these files would impact a system.
Example: How big the journal receiver will get, how many entries, CPU
overhead, should the objects be in the same library, etc. Was not sure if
there is a way to determine this or not.
No answers, but some questions that might help guide your thoughts.
I need to start to journal
Why? The point of this question is to determine if you are keeping
before and after images or just after.
The journal receiver will be twice as large if you keep both images.
Countering that, you may not need to keep those journal receivers more
than one backup cycle. In fact, you can let the system manage the
receivers, and not keep hardly any at all on disk. Depending on why you
are journalling...
How many entries? That depends on the applications that do updates. If
you have a batch type application that touches every record once an
hour, then you'll have (total file records * 24) entries per day. If
you only touch 100 records a day, you'll have 100 journal entries per day.
CPU overhead... hm, not sure I'd worry about that unless you're
journalling a multi-million row table that you touch every row
continuously all day long. The journalling overhead on our system is
very, very small even for the mega files.
I keep my journals & receivers in the same library. I only have one ASP
here; not a big, complicated configuration. I find it easy to keep all
the related pieces / parts together. Helps me maintain what little
sanity I might have left.
There are lots of ways to go about this, and there probably isn't a
single 'right' way. I've stumbled my way along and haven't seen any
real hit on either disk or CPU. The list members with more complicated
setups have seen what constitutes some 'wrong' ways. It will be
interesting to hear their opinions.
--buck
midrange.com
