[VoiceOps] "...but it works with Packet8"
abalashov at evaristesys.com
Mon Aug 23 11:44:59 EDT 2010
On 08/23/2010 11:09 AM, Hiers, David wrote:
> Packet8 is clearly big enough to write their own SIP stack for
> both the client and core devices. Don't know if they did, but
> I would not dismiss the possibility.
This is extremely unlikely, unless they inherited one via licensing
from an acquisition.
Developing a SIP stack is a surprisingly capital-intensive endeavour,
at least, when it comes to working out interop issues and bugs, as
well as natural race conditions arising from a literal interpretation
of RFC 3261. All that testing and R&D is hard to afford. That's why
the only good SIP stacks have been around for at least ten years
(though, from this it does not follow that just because a SIP stack
has been around for ten years means it's any good).
Generally, the only ones with the resources to do it are major
softswitch and/or SBC vendors, not service providers, because of the
level of engineering and software development core competency
required. For service providers development may be important, but is
ultimately rather ancillary to their principal business functions,
especially at the protocol stack level. Successful ITSPs like Packet8
are ultimately sales/marketing/fulfillment-dominated machines, first
and foremost, above all else.
Even then, much commercial gear today licenses the Radvision or
Aricent stack. I know Cisco has their own, as does Siemens and did
CopperCom, and certainly, Alcatel-Lucent. But there aren't that many
of them, ultimately; well-worn commercially viable SIP stacks are a
small family. Venturing outside of that family is where a lot of
problems are found.
Alex Balashov - Principal
Evariste Systems LLC
1170 Peachtree Street
12th Floor, Suite 1200
Atlanta, GA 30309
More information about the VoiceOps