Q: Performance, ZODB3, etc. for an Open Source Projekt (Tracking System)
Hi, some times ago I asked for some advice on implementing a problem/request tracking system. Back then I received a very, very good and stimulating answer. And I started to design the system using some ZClasses, a ZCatalog based search interface and everything else Zope provides for a document based tracking system. This went all very fine and the performance was ok for the first take of the project. So I made the "biggest" mistake in my life - I did a presentation and so on for my boss. He was very impressed and thought that this could be the tool of choice to replace all the various tracking and calendar applications we use in our company. Ok, I thought for round about 100 users Zope on a big machine would be ok and started to implement a first prototype. Well, today my boss discovered, that a lot of ISP offer such a thing for their customers. That means all our customers can use the system for personal Todomanager, calendar application and so on. Various people and commercial customers should be using it as a basis for a collaborative task manager and so on. Ok, our sales is keen on offering such a thingy and this is basically ok. But I, the poor guy who has to develope this thing, have some questions and doubts. Well, I think I should outline the features. The system is a ... - job/request/task/problem tracking system with follow ups - collaborative workgroups should be possible - calendar application (yes, ZClient cron scritps will be required) - users should be able to create subfolders and filters for incoming tasks - multilingual!!!!!!!!!! Is clear what I am about to do? Well, some people might know PTS or Keepstone. I have to develope a better version of these systems or something that fits our needs more then these systems do. Ok, this looks as a Zope application of choice and yes I will use Zope in any case, but .... - What happens if round about 3000 users use the system for work and appointment? This is means round about 6000 to 9000 new documents (ZClasses with ZCatalog support) per day. - I will face 1500 concurrent users, which will access quite a complex Zope application. - How safe is ZODB3 in such an environment? A lot of transactions will happen on this system!!!! Users, ZClient scripts, etc. pp. - Is it possible to split/multiplex a Zopeserver? - Has anybody experience with synchronizing/replacating several Zopeservers? Well I basically like the idea of a pure Zope solution, and I thought that it is perfect for round about 100 users which create roundabout 500-2000 new documents per day. But with this requirements and user number I have some problems to think of a pure Zope solution. I would prefer a Zope frontend for a SQL database. Or am I wrong with my assumption and Zope is much more powerful, then I believe it to be? I can think of a farm of Zopeservers which access a single SQL database server or so, the Zoepservers update the application by retrieving it from a master Zope server and so on. This is something I can imagine. But what about a pure Zope solution? Yes, some might think. Why doesn't this guy go and buy consulting contract from Zope company. Well, the problem is that my company will offer this service as a goody for our customers for free. So no immediate money will be created by the project. And we will release the complete application and so on as an Open Source application, cause we heavily depend on Open Source software as an ISP and so we can give something back to community. If people request support, we will sell support contracts, but we won't charge money for the product or the documentation. So I try to get my answers through the usual Open Source channels. ;-))) Best regards, Oliver -- Oliver Andrich, KEVAG Telekom GmbH, Schlossstrasse 42, D-56068 Koblenz Telefon: 0261-3921027 / Fax: 0261-3921033 / Web: http://rhein-zeitung.de
participants (1)
-
Oliver Andrich