Islands in the Stream : The Mesh WAN
VoIP relies on one thing: the reliability of the internet to deliver a sustained throughput without having to lay an end-to-end physical pipe (or negotiate an equivalent guarantee). Its the concept of an ad-hoc Best-Fit Commited Information Rate Its working. Cable providers and Telcos are flooding clients with tempting service offerings and ever-increasing bandwidth: Vonage, bundled wireless, VoD, FIOS, MegaModemMachX etc. Massive effort is underway to uplift the capacities and capabilities at the core and last mile. One big scramble to “own” the consumer and dominate the market. Peering hotels are spawning rapidly at perceived critical junctures to reinforce the nebulous backbone. In reality its just more Telco “land-grabs” to secure market share and penetration.
In the corporate world, theres a tremendous effort made to distribute the workload and wide-area-network peering to support datacenter failover, balancing and disaster-recovery. One minute down, one delay, can cost millions. Concentrators, in-band compression, wide-area file systems have all been invented to distribute data and load. Most shops create regional, national and continental hubs. Traffic is only ever routed to the optimal hub in relation to that locations peers. Ideally, you structure value-added services to that hub to gain a better ROI – regional directories, mailservers, storage-center, helpcenters etc and all with dedicated pipe: Ts, ATM, ISDN, Satellite. Its a microcosm of the Internet, a great big Hub and Spoke end-to-end pipe. Unlike the internet, the financial commitment and reliability are of paramount concern to both the client and vendor, its really complex to switch from one WAN carrier to another and sporadic outages affect both parties financially.
I argue that sometimes the Best-Fit is good enough. VoIP proves the current generation Internet can contain a robust infrastructure, a reliable model to which the legions of P2P adherents can attest and it doesn’t require an expensive outlay. The wonder of an adaptive routing structure has primarily been confined
to just that, routing and its not enough, we need a mesh WAN or an
“Islands in the Stream” model where each end-point also exists as a
self-contained hive. Islands or Hives should collaborate on sharing the infrastucture optimally and most importantly, none are vulnerable to failures in artificial junctures, the regional hubs of the 90’s WAN model. Forego the expensive carrier-model WAN and subscribe to regional equivalents, route your infrastructure over the internet! Metropolitan networks are cheap in comparison to dedicated lines, run 2 (or more) connections into your hive! CPU-cycles are also cheap, encrypt the hell out of the traffic and build dedicated “bridges” or svc/dedicated channel-failover agreements into the vendor agreement. The best thing is your bottom-line will greatly benefit from regional competition in hot-markets.
The only worry then comes in monopolising/monetizing the govermental-unregulated Cisco over the lobby-heavy and theoretically regulated AT&T? Thats another article…
