r3 - 13 Nov 2003 - 10:40:00 - KaitlinDuckSherwoodYou are here: OSAF >  Projects Web  >  CodeInfrastructureIssues? > WhyNotZodb

Why is Chandler not using ZODB?

Bill Seitz:

Are there any references online (wiki pages, list archives) discussing the decision to eliminate ZODB, and a short explanation of what's being used instead? (Or maybe there are multiple non-ZODB back-ends possible, all hidden behind RAP...)


Andi Vajda:

The reasons are multiple but none are exactly a ZODB show stopper. Ultimately it was a decision I made based on the constraints of the project.

  • ZODB is a python object database. While Chandler is written in python, I wanted its data model to be independant of any language in particular and needed it therefore to be more constrained than python's.

  • By making the data model more constrained and by controlling it there are many things that are made easier, in particular queries, interactions with RDF, bi-directional referential integrity, access control, garbage collection, etc...

  • There are also a number of requirements the Chandler designers created for the repository that are made easier to implement if we control the actual representation of the data and constrain the data model, such as replication, remote access, shared access, threaded transaction model, etc...

  • When I started working on this project in May, the repository was late, very late, and the project was stalled because of that. I felt I could get something usable for the project to resume much faster if I started a data model implementation from scratch and persisted it using Sleepycat's Berkeley dbxml and Berkeley db.


Today, the Chandler repository is not really so much an object database as an item XML database combined with large collections of references directly stored in Berkeley DB. Chandler persistence is grafted on top of whatever language binding is chosen to operate it (currently only Python is supported). By that I mean that a programmer's python instances are never persisted in the Chandler repository, only specific attribute values the programmer intended for Chandler.

The trade-off for these decisions and design choices is a somewhat steeper learning curve for programmers expecting a real object database like ZODB. My hope is that this trade-off is well worth the gains.

-- HeikkiToivonen - 05 Nov 2003


The first point you have written here really matters, but probably you personally feel better with bdb than zodb?

Switching from an unconstrained db to a really low level one... seems quite strange to me...but I'll see later if your is good decision.

-- AzaZel - 11 Nov 2003


Someone who wishes to remain anonymous sent me (Ducky) email that said:

I developed and used a Zope site for two years. The impression I had of Zope was as a fundamentally broken tool. Please note that my experiences deal with an older version of Zope (ca. 2000-2001; I have a poor sense for exact dates) and that I rarely dealt with ZODB as a separate entity. Also, some issues listed under "Zope" also belong to "ZODB" and vice versa. Please also note that Zope made it difficult for me to deal directly with ZODB. Additionally was the problem of having my Zope server provided by a zope-hosting service, so I didn't have complete and unrestricted access to the server. Therefore your mileage will vary on these points.

Zope weaknesses:

  • Web-based editing is as fun as sorting through 100,000 email messages using Yahoo mail.
  • Nearly non-existent session management.
  • DTML is evil. It completely negates the advantages of using Python as a language. For example, only a small number of Python symbols are available for use by the developer and those are set by the Zope designers.
  • Using Python functions requires a large amount of coding overhead, similar to using Java. Again limiting the expressiveness and conciseness of Python.
  • Documentation is horrific. What hasn't been documented is criminal. What has been documented is worse.
  • No test suite capabilities for Zope objects.
  • Search capabilities were primitive. No search and replace. Search itself was limited to simple word search. No searching by project.

ZODB weaknesses:

  • Everything is kept within the database, making them inaccessible to the tools and scrutiny Unix programmers have come to know and respect from file system-based applications. No grep, no sed, no visual inspection. Note that this is an area I hope OSAF keeps in mind when choosing data storage formats as it's a mistake that could be repeated using other storage toolkits. People and enterprises will want alternative methods for viewing their data. This is why the mbox file format has been so persistent for so long. In other words, if you use a non-readable file format you'd better be prepared to also supply your own command line tools for doing greplike things.
  • Primitive search capabilities.
  • Object indices must reside within application code itself.
  • Badly documented (not quite as bad as the main app, but still pretty grisly).
  • Prone to corruption. This happened to me several times so I was left with the impression that ZODB was fragile. Not much in the way of recovery tools (at least then).

Apocryphal point regarding ZODB which I myself haven't tested:

  • Legend has it that ZODB performs well in an environment where there are few writes vs. many reads, but performance suffers under heavier write loads. That's typical of most web-based environments. Indeed I believe it was smart for the database to be tuned for that kind of usage pattern (my own web app followed this pattern). But Chandler will be much closer to 1:1 read/writes and you will want an environment which is amenable to your own tuning.

-- as told to DuckySherwood - 13 Nov 2003


View topic | Edit |  | WYSIWYG | Attach | Printable | Raw View | Backlinks: Web, All Webs | History: r7 |r5 < r4 < r3 < r2 | More topic actions...
 
Open Source Applications Foundation
Except where otherwise noted, this site and its content are licensed by OSAF under an Creative Commons License, Attribution Only 3.0.
See list of page contributors for attributions.