There's a pretty simple rule to follow here: - If your working set fits in memory, pull it in to memory. - If your working set does not fit in to memory, or you require concurrent working sets which will cumulatively not fit in memory, use streams.
It is *always* faster to do transforms and manipulation in memory when the set fits in to memory but it is common for the data not to fit, especially if you're doing anything concurrently. -Mikeal On Apr 7, 2014, at 6:19PM, Chad Engler <[email protected]> wrote: > Benefits of using async streams is not so much raw speed, as ability to do > multiple things "at once". If you were doing this task to 10,000 files, your > methods would process many of those files in parallel; and the non-streaming > version would do them serially. Similarly if you are operating on one large > file, you will save memory by streaming it instead of loading it all and > parsing at once. > > -Chad > > From: [email protected] [mailto:[email protected]] On Behalf Of > Dan Schmidt > Sent: Sunday, April 06, 2014 5:35 PM > To: [email protected] > Subject: Re: [nodejs] Why is my streaming word counter so much slower than a > non streaming version? > > Thanks for the insights. I had tried using much larger text files to compare > the two, but wasn't considering that the synchronous will always beat it, so > long as there is enough memory to read in the file. A better comparison would > be to set them both up as web servers and send multiple requests. > > I did try out combining the processing steps into a single transform to see > the difference and it's performance numbers are roughly the same as the > synchronous version. > > > On Sun, Apr 6, 2014 at 1:57 AM, Ryan Graham <[email protected]> wrote: > On Sat, Apr 5, 2014 at 10:01 PM, Dan Schmidt <[email protected]> > wrote: > So, I'm trying to use streams more and better understand how to use them > properly. > > There was this little contest over at Treehouse to write an app that would > count the words in a text file after being filtered from a certain set of > words in another file. > > Myself and one other person chose to write our apps in node. > > I wrote mine using streams and they wrote theirs without. I have used the > time command to compare the two. Mine is consistently about 5 times slower. > > Link to mine: https://gist.github.com/DanSchmidt/10000777 > Link to their's: https://gist.github.com/pleary/9967501#file-waldo-js > > I could understand if some of my transform and filtering logic was perhaps > less performant than another's, but 5 times slower seems to suggest I'm > probably doing something wrong with the streams. > > Your version is a lot more friendly to the event loop to the point that the > same process could probably server a decent number of http requests if you > threw an http.createServer in there somewhere. > > The faster one, on the other hand, will completely block the event loop after > the second async call to request() finishes. > > So I wouldn't say you're doing anything wrong with the streams, it's just an > extreme use of them. If the dataset is large enough to outweigh the setup > overhead, you could try wrapping some of the pipeline steps in separate > processes. > > ~Ryan > > -- > http://twitter.com/rmgraham > -- > -- > Job Board: http://jobs.nodejs.org/ > Posting guidelines: > https://github.com/joyent/node/wiki/Mailing-List-Posting-Guidelines > You received this message because you are subscribed to the Google > Groups "nodejs" group. > To post to this group, send email to [email protected] > To unsubscribe from this group, send email to > [email protected] > For more options, visit this group at > http://groups.google.com/group/nodejs?hl=en?hl=en > > --- > You received this message because you are subscribed to a topic in the Google > Groups "nodejs" group. > To unsubscribe from this topic, visit > https://groups.google.com/d/topic/nodejs/0jSkPYaaE6s/unsubscribe. > To unsubscribe from this group and all its topics, send an email to > [email protected]. > For more options, visit https://groups.google.com/d/optout. > > > > -- > Dan Schmidt > @codeswish - Blog > -- > -- > Job Board: http://jobs.nodejs.org/ > Posting guidelines: > https://github.com/joyent/node/wiki/Mailing-List-Posting-Guidelines > You received this message because you are subscribed to the Google > Groups "nodejs" group. > To post to this group, send email to [email protected] > To unsubscribe from this group, send email to > [email protected] > For more options, visit this group at > http://groups.google.com/group/nodejs?hl=en?hl=en > > --- > You received this message because you are subscribed to the Google Groups > "nodejs" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to [email protected]. > For more options, visit https://groups.google.com/d/optout. > > -- > -- > Job Board: http://jobs.nodejs.org/ > Posting guidelines: > https://github.com/joyent/node/wiki/Mailing-List-Posting-Guidelines > You received this message because you are subscribed to the Google > Groups "nodejs" group. > To post to this group, send email to [email protected] > To unsubscribe from this group, send email to > [email protected] > For more options, visit this group at > http://groups.google.com/group/nodejs?hl=en?hl=en > > --- > You received this message because you are subscribed to the Google Groups > "nodejs" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to [email protected]. > For more options, visit https://groups.google.com/d/optout. -- -- Job Board: http://jobs.nodejs.org/ Posting guidelines: https://github.com/joyent/node/wiki/Mailing-List-Posting-Guidelines You received this message because you are subscribed to the Google Groups "nodejs" group. To post to this group, send email to [email protected] To unsubscribe from this group, send email to [email protected] For more options, visit this group at http://groups.google.com/group/nodejs?hl=en?hl=en --- You received this message because you are subscribed to the Google Groups "nodejs" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. For more options, visit https://groups.google.com/d/optout.
