On Thu, Nov 8, 2012 at 12:48 PM, Christopher Pollock <span dir="ltr"><<a href="mailto:chris@ionetworks.com.au" target="_blank">chris@ionetworks.com.au</a>></span> wrote:<br><div class="gmail_quote"><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
Sure, but the other value proposition for peering is also performance. <div>
<br></div><div>Obviously the commercials are a concern as transit prices drop and there's a 'worth it' tipping point, but depending on your content delivery strategy, those IX links can be produce some good results.<br>
<div><br></div></div></blockquote><div><br></div><div>Big +1 here.</div><div><br></div><div>Another key consideration with peering links is having multiple _decent_ paths to other domestic networks. In the last year or so there have been numerous occasions where a PIPE switch has failed in one DC or another. I'd much rather have that traffic shift over to another IX (even if it's in a neighbouring state) rather than one or more transit providers, quite often with silly routing.</div>
<div><br></div><div>Likewise, I'm a big fan of having multiple IX operators. Regardless of your opinion of PIPE, it simply makes good sense to try to spread out your risk across multiple sites, IXes and the like - that way (especially in the case of errors in automated configurations or the like being pushed out), you have a much better chance of maintaining performance (bandwidth, latency, etc).</div>
<div><br></div><div>Sam</div></div>