Sure, but when one observes the default vendor position is front to back airflow, if one then applies logic, you can conclude back to front is deployed as a cost cutting measure sans structured cabling.
Kind regards Paul Wilkins On 4 October 2017 at 16:10, Jay Dixon <[email protected]> wrote: > I think Sam's point was that the original email/question was asking purely > about direction front or back, not whether you use TOR switches or > structured cabling back to a central point :) > > On Wed, Oct 4, 2017 at 4:06 PM, Paul Wilkins <[email protected]> > wrote: > >> Sam, >> In an SP environment, you may well have whole rows dedicated to a single >> service - email, or web say. In the rack itself, you'll have web_node_5007, >> web_node_5008 etc. >> >> In the enterprise, you'll have a few email blades, internal web, external >> web, next to a bunch of file and print etc etc etc. These then likely are >> all on different firewall interfaces/firewalls in different zones requiring >> different routing and security. >> >> Kind regards >> >> Paul Wilkins >> >> On 4 October 2017 at 15:41, Sam Silvester <[email protected]> >> wrote: >> >>> On Wed, Oct 4, 2017 at 2:38 PM, Paul Wilkins <[email protected]> >>> wrote: >>> >>>> Because SPs have the luxury to not use structured cabling, due to scale >>>> where all switch ports share a common configuration, so there's no need for >>>> a patch panel, just patch direct to the switch, whereas in enterprise, >>>> inadvertent swapping of ports leads to P1s, hence, structured cabling is >>>> fairly ubiquitous. >>>> >>>> >>> I'm not sure I understand. >>> >>> We're talking about "ToR" switching in this thread i.e. switches share a >>> rack with the servers in question. >>> >>> In both front and rear mounted switches, I'd assume all cables go direct >>> from the server to the switch. If that's not what you mean, can you perhaps >>> share what kind of cabling arrangement you've come across? I'd be >>> interested in how it would work and what logic would go into such a >>> decision. >>> >>> I'm also not sure how such a distinction between enterprise and SP would >>> change anything. SPs would still have a mapping of server to port, it's not >>> like just any server / cable goes into any old port and swapping them to a >>> new/different arrangement during a switch change wouldn't matter...or am I >>> making an incorrect assumption there? >>> >>> Cheers, >>> >>> Sam >>> >>> >>> >> >> _______________________________________________ >> AusNOG mailing list >> [email protected] >> http://lists.ausnog.net/mailman/listinfo/ausnog >> >> >
_______________________________________________ AusNOG mailing list [email protected] http://lists.ausnog.net/mailman/listinfo/ausnog
