[asterisk-users] Distributed FAX - How to best complement asterisk ?

[Ex Vito]

  Hello all,

  I'd like to thank everyone's input which I'll sumarize and comment on
  bellow.

  As in all complex solutions, there are no quick answers and no 100%
  correct solutions. There are trade-offs to be made among
  very different possiblities... Of course, the purpose of my original
  post was exactly get some feedback on what I initially designed
  and to widen my perspective on the particular subject by hearing
  different approaches to the problem.

  It's been great ! :-)

  For those interested, here is the summary:

  1. from Mojo with Horan & Company, LLC

  sheet-fed PDF scanners -> desktop PDF -> print to HylaFAX

  good:
  - nice idea, makes use of centralized HylaFAX server
  bad:
  - needs investment in replacing current equipment
  not sure if:
  - FAX users are PC savvy
  - there is a PC near every FAX


  2. from Andreas van dem Helge

  suggests using a T.38 fax provider

  good:
  - would offload the gatewaying to a provider
  need to know:
  - whether T.38 is effectively solid under such
  scenario (see last comment, below)

  he also comments:
  - no success with callweaver T.38 gateway with some betas

  (answer to his question: the channel banks allow for the
  connection of analog FAX machines to the asterisk servers
  via PRI)

  - then says the topology I presented has too
  many PRIs:

    PSTN <--PRI--> ast 1.2 <--PRI--> AS5300 <--SIP--> T.38 ATA

  he suggests something I don't quite understand
  (are these three "parallel" flows ? or does it represent one PRI
   going to a single AS5300 which would deliver the calls to T.38 ATAs
   or asterisk based on DDI ? what's the difference between the last
   two lines, can the AS5300 talk SIP/T.38 directly to an ATA without
   a SIP proxy ?):

    PSTN <--PRI--> AS5300 <--SIP--> ast 1.2
    PSTN <--PRI--> AS5300 <--SIP--> ast 1.4 <--SIP--> T.38 ATA
    PSTN <--PRI--> AS5300 <--SIP--> T.38 ATA


  3. from Olivier

  shares information he got from Cantata where T.38
  requires good levels of QoS

  my comment:

  I though T.38 was created to bypass those types of technical hurdles
  -- interesting ! (as I'll note below, Steve Underwood helps clarifying this
  notion)


  4. from Phillip von Klitzing

  suggests that some bigger MFC printer/copy/fax combos
  can do FAX via SMTP

  good:
  - great, if it's over SMTP it'll work
  bad:
  - small offices won't justify such a big investment (I used
  to work for HP, I know how much those beasts can cost!) ;-)

  ...unless anyone's aware of a small FAX machine that can
  do SMTP ! (btw, there are some sheet-fed network scanners
  that can do SMTP -- see first comment)

  he also recalls an important issue:
  "are you sure you want to rely 100% on IP only in your sattelite
  offices ? It might be wise to have 1 (analog?) line installed anway

  great point -- this has always been a possibility in the back of
  my mind... the only thing we'd loose in a setup where the remote
  office FAXes are directly attached to local analog lines is the
  ability to do integrated CDR processing for those FAX usages


  5. from Benny Amorsen

  reminds that those big MFC boxes require the "fax as email address"
  for sending -- maybe too complex in day to day usage ? how tech
  savvy are the users ?

  another good point -- apart from their cost, in terms of usability, they
  might come short... or be too complex for someone with basic FAX
  machine abilities


  6. from Steve Underwood

  reminds that T.37 (store and forward instead of realtime)
  is the answer to reliability... T.38 isn't all that robust, it just isn't
  as awful as FAX over VoIP

  he then concludes "In a sane world all FAX would have
  been T.37 from a few months after the spec was released"

  great info -- so, where is the T.37 compliant equipment ?
  (gateways, ATAs, FAX machines ?)


  Again, thanks a lot for the feedback (keep those posts coming!).
  Meanwhile I'll move on to further investigate some of the alternatives
  you proposed.

  Cheers,
--
  exvito