Please login or register to participate.
chist_alex Feb 17, 2010 06:02 PM
Can you tell some examples of installations of your clients in numbers - the heaviest data.fs file, the biggest amount of objects, the biggest amoint of users AND how does cynin works in those conditions? How is its performance?
Can you tell any limitations you know which slow down cynin?
Replies (5)
chist_alex Feb 20, 2010 12:32 AM
I know limitations of Zope. Just very interesting how works cynin itself in different conditions.
acsr Feb 21, 2010 08:59 PM
If you know the limitations of zope, you should look at the limitations of Plone! is in my opinion a very cool incarnation of Plone that comes up with: perfect skin, perfect productivity preconfiguration, cool support possibilities. But it is still Plone when it comes to content and users. There are known sites with over 50.000 user accounts. Data.fs files with very huge size. There is a limitation if it comes to lot of users authoring at the same time. So you have to deliver the infrastructure for this (ZEO, ZEO-Raid, NGINX i.e.). There where extreme load tests during the naples plone conference sprint 2007 done with Plone 3. I do not know the links where they were published. Search for them! There is a good chance that exactly matches the limitations of the underlying Plone 3 installation architecture (by design!)
chist_alex Feb 22, 2010 01:50 PM
Plone is Okay. Im sure. I'm asking about CYNIN. Because of a lot of javascript which loads and sorts tonns of data.
chist_alex Mar 02, 2010 09:00 PM
don't have time to answer or bad question?
dhiraj Mar 10, 2010 11:49 PM
Not a bad question.

The short answer to all your questions in your initial post which start with "Can you?" : No.

Long answer:
Data.fs file size is not linearly proportional to the amount of data + revisions you have. It's also how many files you have, how much undo history you're needlessly storing, and so on. Like acsr mentioned, is based heavily on Plone and Zope and architecturally has similar performance as that when you look at it from an infrastructure requirements perspective. Clear numbers such as what you're looking for cannot be given because they not only depend on, but are pretty much *governed by* what kind of infrastructure you have and by what kind of load you intend to submit to it. This is not a problem with, but rather with the way you're looking for a solution. You should not ask "What limits can reach?", but rather "What do I have to do make reach the limits I need it to."

To answer your questions with some semblance of sanity, this is what we have to do:

1. First, we have to ask *you* the figures you want to achieve: How many users you have in all, and how many will be concurrently accessing the system for each type/segment of user? How many file objects do you have in each size bracket (0-5Mb, 5-25Mb, 25-100Mb, 100Mb-1Gb, 1Gb+). How many and what types of objects do you primarily have (Wiki pages, Discussions, images, audio, video, etc.). What is the growth ratio for your total and concurrent users as well as the above content parameters?
2. Then we get to build a test system that we submit to the kind of load you're looking to achieve right now (of users and data), and we factor in a percent capacity buffer for your usage timeline for growth.
3. Based on above we offer infrastructure recommendations and have one of our partners go ahead and setup something that will accommodate your requirements while working within your budget constraints.

Platform performance is something that we and the whole Plone community strives to improve on with every revision that we put in at every level of infrastructure right from architecture to end UI component widget. Plone 4 is achieving new frontiers in this regard, from what I'm hearing on twitter, IRC and so on, so you should probably hold on for that to "blow your worries away" if that's what you're waiting to hear for before jumping on to using I'd say stop waiting to try, and try now. Life is short. ;)

We do use a lot of javascript with, but that's not the performance killer from a server side architecture perspective. That war begins and is for the most part fought out right on your client-side browser.

The server by it's very nature is a "resource hungry server application" and will gobble up all the server resources you give it (and then some!) if you have sufficient load at the other end.

What will help you if you really want to go down the route of setting up your own infrastructure without professional assistance?
Skills: Good knowledge of Linux tech. on the server side with regards to load balancing, software clustering and production server hosting.
Server: Fast disk I/O. Lots of RAM. CPU Cores.