Bandwidth Savings

Fletcher Kittredge fkittred at gwi.net
Wed Jan 11 19:36:52 UTC 2017


The problem with the local cache[s] is the bandwidth cost of populating the
cache and keeping it coherent can be greater than the bandwidth saved. From
your description, I would expect this to be the case so a local cache will
not help. Rule of thumb is if your downstream traffic is not at least
3gb/sec, you won't see a win from a cache. This problem can be mitigated if
you can find other large bandwidth consumers on the island and partner to
share a cache. Examples of potential partners would be your competitors,
universities, government organisations, etc.  The savings can be
significant.

If there is a local peering point on the islands, this would be the best
place for shared caches. Sharing caches via an existing non-profit peering
organization or having a non-profit, educational organization, or the
government take the lead can lower the suspicion barrier and result in more
sign-ups.


On Wed, Jan 11, 2017 at 1:58 PM, <Valdis.Kletnieks at vt.edu> wrote:

> On Tue, 10 Jan 2017 23:08:45 -0500, Keenan Singh said:
>
> > do have a Layer 2 Circuit between the Island and Miami, I am seeing there
> > are WAN Accelerators where they would put a Server on either end and sort
> > of Compress and decompress the Traffic before it goes over the Layer 2, I
> > have never used this before, has any one here used anything like this,
> what
>
> Those will probably not help a lot with https: data, as a properly
> encrypted
> stream is very close to random bits and thus not very compressible.
>
> As others have noted, your best chances are getting content providers to
> give
> you a local cache of their most popular content.
>



-- 
Fletcher Kittredge
GWI
207-602-1134
www.gwi.net



More information about the NANOG mailing list