Posted:
28-May-2007 07:40 I am investigating various approaches for developing a web-app that (if successful) will need to scale to millions of page hits per day and up. Has NetKernel been used in such an environment? I understand that the concepts upon which it is based are specifically intended to enable scaling. But I'm wondering if NetKernel (as an implementation of those concepts) has been pounded on enough to verify that it has the durability and internal performance to handle such loads. I'm imagining that NetKernel would be on back-end systems, but I'm also wondering if anyone has tried to use it on scaled-out front-end systems. My rough idea is something like this: back-end cache front-end -------- ----- ---------
Netkern-1----- -----httpd-1 | | Netkern-2----------Squid------------httpd-2 | | NetKern-N----- -----httpd-N |
In practice there would be many more front-ends than back-ends. The primary idea behind squid is to consolidate caches, but it also has many nice cache specific features. The back-end NetKern servers would not all be identical (although some might be load balanced clones) and sit atop various resources like RDBMS, static files, external data, etc. Anyway, I've been pondering this all day and wanted some feedback since it's rather different from what I've done previously. Thanks, -Brad |