cxf-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Daniel Kulp <dk...@apache.org>
Subject Logging/cachedoutputstream defaults for 3.0....
Date Mon, 24 Mar 2014 18:06:22 GMT

I’d like some thoughts on changing a couple of defaults for 3.0.

We currently have the default size for when the CachedOutputStream flips from in-memory to
using a file store set at 64K.   This was chosen as the default a long time ago.   

In the Logging interceptors we, by default, only log up to the first 100K of the message.
  

Up until this past Friday, the LoggingInterceptors would cache the entire message so flipping
to a file during logging wasn’t a serious issue as we’d already have to flip to the file
to cache the whole message.  However, on Firday, I changed the logging stuff to only cache
the “limit” and then stream the rest.   Thus, we’re now in a situation where for messages
more than 64K, we’ll create a 100K file for the cache, log it, then the 100K file gets fairly
quickly deleted as the message is processed/consumed.   


So, the question is:   do those defaults still make sense?    

There are two thoughts:
1) 64K is relatively small nowadays.   That value has been there for a VERY VERY long time,
but in the mean time, people’s VM’s are now much larger.   Increasing this to 128K or
so may make some sense.   That could speed up some other usage of the CachedOutputStream as
well.    Downside is increased memory usage.

2) 100K logging is a lot.  Seriously.   I’m not sure how much “extra” information we
get from logging 100K compared to say 32K or even smaller.


What are peoples thoughts about changing those value for 3.0?   I think the first one makes
a lot of sense now.   If 64K was OK 5 years ok, I would think 128K would be OK now.   Not
really sure on the second one.   Thoughts?


-- 
Daniel Kulp
dkulp@apache.org - http://dankulp.com/blog
Talend Community Coder - http://coders.talend.com


Mime
View raw message