VoIP Developer

TMCnet - The World's Largest Communications and Technology Community
New Coverage :  Asterisk  |  Call Recording  |  SIP Trunking  |  Fax Software  |  Load Balancer  |  PBX  |  CTIA  |  INTEROP  |  Small Cells
VoIP Developer Feature Article
> VoIP Developer home

David Duffett

[See Other Articles by David Duffett]

 

[August 26, 2005]

The Voice of IP
Friendly Faces at the VoIP Developer Confab

BY DAVID DUFFETT


SAN FRANCISCO -- Earlier this month, this great event saw record numbers of conference delegates packed into different sessions, keen to find out more about the latest developments in this fast moving sector.

VoIP Developer Conference

The photos above are of the nice people who attended one of the sessions I ran. There are more photos of other sessions on the TMCnet Web site.

The overwhelming feeling I got is that VoIP has moved a long way forward in so many different dimensions. Last year, the emphasis seemed to be on actually getting IP connectivity; this year, the focus has shifted to differentiated service offerings and other aspects more concerned with functionality.

HMP, Host Media Processing
A good part of the proceedings was devoted to HMP. Various offerings were on display, with differing providers claiming highest channel densities, most functionality, etc. One of the things that slightly irritated me was the huge number of times Moores Law got quoted (incorrectly most of the time), and referred to with words like riding the Moore's law curve, (which is difficult because it's not a curve...it's a linear relationship.)

Lets get this Moores Law thing straight:


(mrz l)
   (n.)
  1. The observation made in 1965 by Gordon Moore, co-founder of Intel, that the number of transistors per square inch on integrated circuits had doubled every year since the integrated circuit was invented. Moore predicted that this trend would continue for the foreseeable future.
  2. In subsequent years, the pace slowed down a bit, but data density has doubled approximately every 18 months, and this is the current definition of Moore's Law, which Moore himself has blessed. Most experts, including Moore himself, expect Moore's Law to hold for at least another two decades.
SOURCE: Webopedia

So, the current definition is that the number of transistors per square inch on an integrated circuit doubles every 18 months. It is generally taken that this means processing power will double every 18 months as a result, and this is a fair assumption.

Notice that Mr. Moore did not apply his law to just CPU chips but to integrated circuits generally. Thus, DSP technology also benefits from Moores law.

Lastly, we must remember that Moores law does not operate in a vacuum, and while it may be true that processing power doubles every 18 months, it is also true that the demands on this processing power are also increasing across time although the rate of increase is somewhat difficult to determine. We cannot, therefore, assume that we are going to be able get twice as much done in everyday applications every 18 months.

Enter Duffetts Law.
Duffett's Law is a relatively simple derivative. It states:


"When you get really ticked off by people constantly referring to Moores law, simply remind them that the demand on available processing power is also increasing at the same time (but at a less easily determinable rate).

SOURCE: David Duffett


Duffetts law can be easily proven by observing that your current desktop or laptop PC does not boot up any faster (it may even be slower) than the PC you were using years ago. Why? Because now more stuff is going on. Windows XP is doing more than Windows 98 did. Things like anti-virus, anti-spyware and firewall software were not such desperate needs back then. Ask a really old tech-head how quickly his DOS machine booted up and his answer may surprise you. (IMPORTANT NOTE: Be careful when talking to such individuals as they may keep you talking longer than you really want so plan an exit strategy ahead of time.)

So Whats the Real Deal with HMP?

I thought that the keynote panel on HMP at the VoIP Developer Conference gave very fair and balanced views on why and when to use this kind of technology.

One of the most poignant observations made -- modesty prevents me from identifying the guy that made it -- was that the choice to go HMP, or not, must be made on more than just the functionality requirements of the application. It must take into account the whole raison d'tre of the complete solution.

As an example, if there was a need to build an IP-based solution for a maximum of 120 specially trained medical staff to receive direction in the event of an emergency, one could argue that an HMP solution was just right in that it would be capable of terminating 120 calls and playing out the appropriate messages.

An HMP solution, however, is not totally deterministic; that is, one cannot predict the exact way it will respond in all different scenarios.

Imagine if those 120 channels were tested using G.711 or G.729 codecs; but then, in its first real-life use, the majority of people called in using the G.723.1 (higher complexity) codec. What would happen? Would the last few channels just not work? Would speech quality degrade for some or all of the calls? The simple answer is that one cannot know until it occurs.

And what happens when you need to double the size of the system, do you just buy another host? Or will the integration of another host demand the re-architecting of the solution?

With a DSP card-based solution, there will usually be a well-defined set of performance data so it can be determined precisely what will happen when the system gets to capacity. If you need to double the size of the system, you buy another card.

The example used may be a little extreme but its to prove a point. In a solution that is not mission-critical, another set of rules apply.

The key message must be to weigh the whole application scenario not just make a decision on the application functionality alone.

Get Away with a Gateway?

Gateways continue to be big business with the need to interface existing TDM-based equipment to new IP telephony infrastructures and networks very much in evidence.

A quick trip around the excellent exhibition floor that accompanied the VoIP Developer confab showed the leap since last year in channel density of VoIP boards with T1/E1 interfaces is quite incredible. These are the components on which gateways are built.

One exhibitor boasted a single PCI card that could act as a gateway between 8 traditional T1s and SIP or H.323 IP telephony. When an IP connection becomes the norm with regular telcos, how many enterprises are going to be crying out for this kind of solution? VoIP developers, take note! This could be your next revenue stream!

If youre a VoIP developer or interested in buying VoIP solutions...

  • Come along to the next INTERNET TELEPHONY Conference & Expo - its in Los Angeles at the LA Convention Center on October 24-27, 2005.
  • I will be there, come along and say hello at the following panel sessions:
    1. VoIP: Where Were Going, Where Weve Been
      Tuesday 10/25/05, 10:1511:00am
    2. The Future of IP Telephony
      Tuesday 10/25/05, 3:45-4:30pm

See you next month, when well ask: Open source and the traditional business model; can they be buddies?

David Duffett
[email protected] 

David Duffett, TMCnet's columnist for "The Voice of IP," is a Chartered Engineer and has been in the Telecoms sector for over 14 years with experience spanning air traffic control communications, wireless local loop, mobile networks and computer telephony. At Aculab for 5 years, David has global responsibility for customer training through the Aculab Academy.

VoIP Developer


TMC LOGO
Technology Marketing Corporation,
2 Trap Falls Road Suite 106, Shelton, CT 06484 USA
Ph: 800-243-6002, 203-852-6800; Fx: 203-866-3326
General comments: tmc@tmcnet.com. Comments about this site: webmaster@tmcnet.com.
About   Contact  Advertise
Technology Marketing Corp. 1997-2024 Copyright. Privacy Policy Sitemap