[Twisted-Python] From prototype to production
Hi, Many thanks to the creators and developers of Twisted. I have been using Twisted to develop a prototype (using an Dell desktop) for an inline DPI server. Needless to say, Twisted made the development time much shorter, and I am very grateful for that. I now want to migrate from prototype to production. I was fully expecting to use an intel IXP server, and having to learn its OS and low-level programming languages. But intel itself told me that they are discontinuing that family, and are recommending the use of multi-core general-purpose servers. If that is the case, then it is tempting to think that I can use the exact same python software I developed for the prototype in the production server. Mind you, for the prototype I was testing three simultaneous users. For production, I am likely to have 5000 simultaneous users per server. Is this thinking correct? Are there any obvious scalability issues that I would encounter if I used python in production? Is it automatically obvious that I need to reprogram my methods in C, or even a lower-level language? Thanks. N. _________________________________________________________________ Stay in touch when you're away with Windows Live Messenger. http://www.windowslive.com/messenger/overview.html?ocid=TXT_TAGLM_WL_messeng...
On 03:38 pm, newmom72@hotmail.com wrote:
Is this thinking correct? Are there any obvious scalability issues that I would encounter if I used python in production? Is it automatically obvious that I need to reprogram my methods in C, or even a lower-level language?
It's not "automatically" obvious, but that doesn't mean that it will work. Develop some benchmarks for your application, see how fast it is, determine whether you need to optimize it, and profile it to discover which parts would benefit from optimization. This strategy is not unique to Twisted, Python, or any other library or language, for that matter. Even if you wrote your whole system in C there is no guarantee that it would magically be fast enough to support your users. In most cases Python is fast enough, and in those cases where it is not, you can usually move just a few functions to C. I have never seen a case where Python was so slow that an entire system needed to be rewritten.
participants (2)
-
glyph@divmod.com
-
nadine