A Fault Tolerant High Transaction SIP Network
VoIP Technology provides some wonderful new advantages over TDM switching architectures. One advantage is the ability to have a customer's CPE equipment be dynamically tied to a group of NSIP switches, which allows calls to be distributed in real-time across that switch set. Realizing such architecture offers both redundancy and throughput advantages that are not possible with circuit-based technologies. The following paper describes how Cyclix created the N-Plus™ redundant architecture.
TODAY'S SIP BASED VoIP
VoIP networks and services have grown way beyond the point of curiosity. In fact, the technology is radically changing the telecommunications industry by attempting to rival and compete with legacy circuit-switched PSTN networks. But, as with any emerging technology, hurdles will present themselves as the use of the technology evolves. With VoIP, a novel approach is now needed to overcome the issues relating to SIP throughput and redundancy - two areas that have become a significant issue. The problems specifically are:
- Bandwidth Heavy - SIP utilizes an abnormally high rate of bandwidth for signaling; i.e. ~15 kbps for each sustained single call per second. That is a lot of data to process for each call!
- Character Clumsy - SIP is somewhat difficult for computers to handle. The protocol is all text, which must be parsed using relatively computational-intensive text manipulation software.
As a result of these issues, today's fastest servers are being challenged when attempting to rival switching speeds of the SS7- ISUP based TDM networks. Thus far, the general solution has been to give end-point SIP devices a hard coded IP address to the SIP carrier's proxy (i.e. the SIP switch). Then, at the point where the SIP carrier's proxy runs out of cycles, and can no longer handle any further load, the SIP carrier asks the end user to use yet another IP address to yet another SIP proxy. This approach is problematic in that it requires the SIP endpoint device to manage its own traffic termination. At the same time, this creates single points of failure for the end user. Cyclix has looked at the situation and found a way to resolve the dilemma by using a clustered architecture.
THE N-PLUS™ MATRIX
At the core of Cyclix's network there are a pair of DNS servers that direct the SIP end point user agents (UA). The Cyclix DNS servers resolve to (what appears to be) a Cyclix proxy address for the UA. In reality, the UA is pointed to something new in SIP space, which Cyclix calls an STP (signaling transfer point).
This mnemonic/name was adopted from SS7 space because Cyclix engineering realized it did something similar to its SS7 counterpart. Specifically, Cyclix STPs (deployed in groups for redundancy) actually decide which one of N proxies in the Cyclix proxy matrix will be used to process a given SIP call for a given SIP endpoint.
The STP functionality allows Cyclix to deploy a matrix of "N" SIP proxies to handle whatever load is required for any given task at hand. There is no limit to the CPS (calls per second) processing power within the Cyclix signaling domain. Furthermore, there is no need for the UA's to hard code a carrier's proxy addresses into their configuration schemas. So, the UA no longer needs to be concerned if a proxy has the capacity to handle its load.
A Fault Tolerant High Transaction SIP Network Using this N-Plus™ approach also eliminates any single point of failure, and realizes redundancy to the nth degree. Cyclix's N-Plus™ Architecture solves some major technical hurdles just now surfacing in the SIP market place.
As mentioned prior, Cyclix has deployed a matrix of proxies, with each call being able to use any of the proxies in that matrix. This presents another issue. Each proxy must now know registration and routing information for all the SIP endpoints on the Cyclix SIP network.
To solve this problem, Cyclix has created the "Synchor" and "Registrar" services. These two services run redundantly in real time, keeping the registration and routing information current and equal across all Cyclix proxies.
Again, Cyclix has realized another industry first in its approach to dealing with the problem of high capacity telephony signaling on an IP network. The ability to be able to synchronize all proxies in real time creates a single virtual switching machine with unlimited call processing potential.
Finally, Cyclix ties this unique N-Plus™ Architecture back to the PSTN, with a wide spectrum of tier-one TDM national and international carriers for TDM network access. This offers advanced routing options for Cyclix customers, regardless of origination or destination of call.
Stepping back and looking at the whole picture, Cyclix is able to fill a critical need for high volume SIP customers by offering:
- High capacity throughput
- PSTN or better redundancy & reliability
- Multiple routes to any global destination
- Lowest cost route to any given destination
- Ease of connection to the PSTN
Acting as a high-capacity, homologated SIP switching point for SIP endpoints, and connecting those endpoints to the world, Cyclix is now the "virtual IP central office" to the world . For further technical information, please contact Cyclix sales: