Greetings, On 3/17/11, Venkatraman S <[email protected]> wrote: > > Would like to know the network level details as much as possible; more than > network level, > i am interested in how the webserver acts with the mpeg file that resides on > the server. >
May I, as usual, intervene here with another layering within the "webserver " and "mpeg file that resides on the server" Now, in a serious media server, clips shouldn't be residing servers. It should be residing on HA storage (SAN,iscsi, NAS, NFS etc.). Now in that scenario, that "know the network level details as much as possible" in itself is an Industry, per se. What I had in mind was IT infrastructure for an HA media server. Typically a pair of physical servers accessing a data store through copper/iscsi/fibre/FC network etc. would get data at around 4gbps into the server boxen which is thrown out by the server boxen which would have many services running within them like iscsi-initiator-d, multipathd, httpd, usual network deamons etc. Now one of them would be, httpd, apache webserver which listens on the VIP provided by your HA hardware. and you can put apache under a cluster managed service an application server like tomcat etc can be. Databases like oracle, postgresql, mysql etc can made active passive configuration. One can have underlying sybsystems like CLVM etc.which can host HA clustered filesystems as GFS, OCFS etc. which makes a lovely base for having a samba/NFS share. Now who mincemeats the URL is httpd and its handlers' headache. Apache (Linux process name: httpd) just knows that so and so file is stored in a certain filepath which usually points to external storage (when seen from within that box) mounted on the some suitable mountpoint on physical servers's filesystem. Now, from UML perspective etc., a little later. HTH Regards, Rajagopal _______________________________________________ ILUGC Mailing List: http://www.ae.iitm.ac.in/mailman/listinfo/ilugc
