Technical

Technical

Hardware and redundancy

EAD operates within two points of presence, Interxion in Vienna and NX Data in Bucharest. Each one of them is linked to two distinct world-level Internet carriers, both of them delivering a balanced service between the two points.

Both Internet carriers are linked to a couple of Cisco routers. The array of equipment behind the routers is utilized at a 50% level of the total capacity, in order to be ready to sustain the whole traffic in case of a hardware failure. This is also aimed at avoiding a major SLA event. This architecture ensures a good level of redundancy.

EAD interconnection technologies consist mainly of VoIP (SIP) interconnections, along with a number of TDM (SS7) ones.

Brief description:

  • Cisco routers with redundant protocols (HSRP) and dynamic routing protocols (BG4, OSPF, etc.);
  • Cisco switches with VLAN and trunking algorithms;
  • Arrays of SS7/IP media controllers and gateways;
  • Session Border Controllers in redundant arrays;
  • IP and TDM switches, class 4;
  • Various alert systems, capable to send mails or SMS to the NOC team.

 

Scalable and cost-cutting architecture

We’ve started to construct our system by following two main ideas: it had to be very scalable and robust. Meanwhile, we have tried to be mindful of costs.

The entry points in the system are represented by a couple of Cisco routers, working in tandem for redundancy reasons, both of them linked to two Internet carriers and managing their own C-classes.

All the voice traffic is directed to an array of servers. The calls are distributed evenly, every server having, at the time being, a number of calls approximately equal with the others. If one server is down (e.g. hardware failure), the others will take the traffic until a spare one is started. In the case of transit capacity expansion, only a new server must be sent to the PoP. The system has a good level of scalability.

The robustness of the system is certified by the fact that we have had no major incidents throughout our activity, despite some hardware failures, Internet carriers out of service or traffic floods.

When it comes to finding a cheap solution, namely when the possibility of expanding transit capacity emerges with a certain number of simultaneous calls, the estimated ROI period is of about a month.

 

Software

Voice industry works with huge amounts of information, thousands of nets, millions of prices, tens of thousands of routing options. Therefore, it is useless to speak about the importance of software systems.

We have chosen to develop our own tailored in-house software system.  

We are using:

  • A robust and scalable database engine;
  • Multithreaded capabilities; this means that the software will know how to use any number of processors that are available on the server machine;
  • Arrays of RAID SSDs and HDs, on different machines, with partitioning architecture and real-time replication, for maximum security and speed;
  • A client-server architecture;
  • An online replication mechanism, in order to permanently have “mirror” database on a secondary/tertiary machine;
  • Failover mechanism and “what-if” scenarios;
  • Almost real-time CDR analysis, which ensures quality monitoring of the calls;
  • Real time alerts on mail or SMS send to the Network Operating Center;
  • Robust mechanism ensuring almost instant removal of a faulty server from the array.

We are aiming towards a strong informational system and we strongly believe that we are going in the right direction.