VoIP Mailing List Archives
Mailing list archives for the VoIP community |
|
| View previous topic :: View next topic |
| Author |
Message |
abalashov at evaristes...
Guest
|
 Posted: Mon Apr 07, 2008 1:01 am Post subject: [asterisk-users] Where is the Digium DS3 card? |
 |
|
Michael Cargile wrote:
| Quote: | Another reason I am sure that Digium has not released a DS3 TDM card is
the fact that asterisk currently cannot handle that many channels. I am
speaking from experience on this. We have build before a predictive
dialer with 16 PRIs. In order to do this and not have audio quality
issues we had to use an 8 core Intel Xeon server with 16 gigs of ram, a
6 drive RAID 10, and two octal echo canceling Sangoma cards. This also
required numerous OS tweaks and dial plan optimizations. The amount of
time spend on this was not worth the final product.
|
I hope some optimisations were made to the ViciDIAL code as well, where
surely the greatest efficiency gains are to be reaped.
| Quote: | In the mean time, if someone really needs to handle that many channels I
would suggest purchasing a DS3 to T1 mux and pass the T1s onto mutliple
Asterisk servers setup in a cluster. In the end you will end up spending
far less money and time setting the system up. I also saw recently at a
trade show a DS3 to SIP converter which might also lower the cost as you
would not need T1 cards. The only issue is that they are a some what new
technology where as DS3 to T1 muxes have been around for some years now
and can be found on ebay for around 700 dollars.
|
Good ISDN gateways with DS3 interfaces that can output a whole DS3 of
VoIP channels are hard to come by. A Cisco AS5400 claims to, and has
the DSP density for it, but between its processing power and its TDM bus
cannot handle much more than about half of that. Of the Cisco media
gateway line, as AS5850 is probably your best bet.
A Lucent TNT Max outfitted with _plethoric_ VFCs might work okay. Apex
too, perhaps. Haven't tried to see how much it can handle when TDM->RTP
translation is required.
The cluster idea you suggest is probably most economical, but it sure
does require a lot of redundant T1 interfaces.
--
Alex Balashov
Evariste Systems
Web : http://www.evaristesys.com/
Tel : (+1) (678) 954-0670
Direct : (+1) (678) 954-0671
Mobile : (+1) (706) 338-8599 |
|
| Back to top |
|
 |
stotaro at totarotechn...
Guest
|
| Posted: Mon Apr 07, 2008 1:16 am Post subject: [asterisk-users] Where is the Digium DS3 card? |
|
|
On Mon, Apr 7, 2008 at 2:01 AM, Alex Balashov <abalashov at evaristesys.com> wrote:
| Quote: | Michael Cargile wrote:
| Quote: | Another reason I am sure that Digium has not released a DS3 TDM card is
the fact that asterisk currently cannot handle that many channels. I am
speaking from experience on this. We have build before a predictive
dialer with 16 PRIs. In order to do this and not have audio quality
issues we had to use an 8 core Intel Xeon server with 16 gigs of ram, a
6 drive RAID 10, and two octal echo canceling Sangoma cards. This also
required numerous OS tweaks and dial plan optimizations. The amount of
time spend on this was not worth the final product.
|
I hope some optimisations were made to the ViciDIAL code as well, where
surely the greatest efficiency gains are to be reaped.
| Quote: | In the mean time, if someone really needs to handle that many channels I
would suggest purchasing a DS3 to T1 mux and pass the T1s onto mutliple
Asterisk servers setup in a cluster. In the end you will end up spending
far less money and time setting the system up. I also saw recently at a
trade show a DS3 to SIP converter which might also lower the cost as you
would not need T1 cards. The only issue is that they are a some what new
technology where as DS3 to T1 muxes have been around for some years now
and can be found on ebay for around 700 dollars.
|
Good ISDN gateways with DS3 interfaces that can output a whole DS3 of
VoIP channels are hard to come by. A Cisco AS5400 claims to, and has
the DSP density for it, but between its processing power and its TDM bus
cannot handle much more than about half of that. Of the Cisco media
gateway line, as AS5850 is probably your best bet.
A Lucent TNT Max outfitted with _plethoric_ VFCs might work okay. Apex
too, perhaps. Haven't tried to see how much it can handle when TDM->RTP
translation is required.
The cluster idea you suggest is probably most economical, but it sure
does require a lot of redundant T1 interfaces.
--
Alex Balashov
Evariste Systems
Web : http://www.evaristesys.com/
Tel : (+1) (678) 954-0670
Direct : (+1) (678) 954-0671
Mobile : (+1) (706) 338-8599
|
A T3 MUXed into 28 T1 PRIs in one, or a few trunk groups inherently
has redundancy. If a box dies, the calls are dropped (unless you are
doing reinvite) and any call backs go right to the
Ts that are not in alarm.
Running stripped down Linux OS boxen with quad port T1 cards and four
or five lines in extensions.conf, no unneeded modules or software
loaded, asterisk 1.2, entries for zaptel and zapata, and a couple
entries in sip.conf builds a worry free solution. Asterisk uptime 2
years, system uptime two years and twenty minutes.
Thanks,
Steve Totaro |
|
| Back to top |
|
|
abalashov at evaristes...
Guest
|
| Posted: Mon Apr 07, 2008 1:46 am Post subject: [asterisk-users] Where is the Digium DS3 card? |
|
|
Steve Totaro wrote:
| Quote: | A T3 MUXed into 28 T1 PRIs in one, or a few trunk groups inherently
has redundancy. If a box dies, the calls are dropped (unless you are
doing reinvite) and any call backs go right to the
Ts that are not in alarm.
|
True - and if you're simply using CT3 as an economical method of getting
say, a dozen T1s into a gateway, that is probably an advantage. But if
that's the case, it would not be cost-effective to shell out extra money
for additional PCs with quad T1 cards just to provide failover in the
event that the primaries fail. 80/20 rule and all that.
The point is that most people that want a DS3 interface really do want
to pump in a DS3's worth of calls, more or less, in which case they
really can't afford to have those DS1s going spare just for redundancy's
sake. And if you are doing substantially less than a DS3's worth of
calls, you probably shouldn't be looking at a DS3 interface to begin
with unless that's just an incredibly lucrative way to get channelised
PRIs in from a vendor - and with typical the cost of UNE DS3 loops vs
T1s, that's not necessarily so.
Secondly, an industrial-grade ISDN media gateway designed for telco
environments (like a Cisco AS, say) isn't going to go down frequently
enough to merit this kind of concern. Don't get me wrong, I am the last
to go on record saying that Cisco voice equipment (or any other) doesn't
fail from time to time -- ha. But, again, 80/20 rule. A PC is much
more likely to fail within the same MTBF.
So yes, a single gateway handling a DS3 can go down. But so can an M13
mux. You've got single points of failure either way.
If one is in the sort of environment where such high availability really
is a concern (typically a telco setting), one probably needs to invest
in a big DACS and redundant, protection-switched DS3 paths (and
protection line cards for them on the DACS side) as well as redundant
gateways, or at least redundant DS3 line cards in the chassis. At that
point of stringent availability, this discussion becomes a wee bit moot
because most likely you would not be using Asterisk and PCs in such a
setting anyway.
| Quote: | Running stripped down Linux OS boxen with quad port T1 cards and four
or five lines in extensions.conf, no unneeded modules or software
loaded, asterisk 1.2, entries for zaptel and zapata, and a couple
entries in sip.conf builds a worry free solution. Asterisk uptime 2
years, system uptime two years and twenty minutes.
|
Yes, but total cost of ownership goes up because you need someone to do
all that, and even so, despite the impressive uptime you mention, PCs do
need a lot more maintenance, upkeep and worry.
With dedicated media gateways, you just plug in, set up and it works.
--
Alex Balashov
Evariste Systems
Web : http://www.evaristesys.com/
Tel : (+1) (678) 954-0670
Direct : (+1) (678) 954-0671
Mobile : (+1) (706) 338-8599 |
|
| Back to top |
|
|
g.stewart at horwits.c...
Guest
|
| Posted: Mon Apr 07, 2008 4:48 am Post subject: [asterisk-users] Where is the Digium DS3 card? |
|
|
On Sun, 6 Apr 2008 17:22:58 -0400, "Jay R. Ashworth" <jra at baylink.com>
wrote:
| Quote: | Yes, I've seen that, and most of its arguments are specious, at best.
They amount to "I am too stupid to find a mail user agent with List
Reply, and too lazy to switch to it".
|
Are there any MUAs (other than Microsoft's pitiful offerings) that do not
observe RFC2369 headers?
--
Godwin Stewart - Horwich IT services |
|
| Back to top |
|
|
stotaro at totarotechn...
Guest
|
| Posted: Mon Apr 07, 2008 6:31 am Post subject: [asterisk-users] Where is the Digium DS3 card? |
|
|
On Mon, Apr 7, 2008 at 2:46 AM, Alex Balashov <abalashov at evaristesys.com> wrote:
| Quote: | Steve Totaro wrote:
| Quote: | A T3 MUXed into 28 T1 PRIs in one, or a few trunk groups inherently
has redundancy. If a box dies, the calls are dropped (unless you are
doing reinvite) and any call backs go right to the
Ts that are not in alarm.
|
True - and if you're simply using CT3 as an economical method of getting
say, a dozen T1s into a gateway, that is probably an advantage. But if
that's the case, it would not be cost-effective to shell out extra money
for additional PCs with quad T1 cards just to provide failover in the
event that the primaries fail. 80/20 rule and all that.
|
7 HP DL320s, RAID 1 with Quad Sangoma. Not a dozen but more than
twice that, 28 T1s. What is your cheaper solution? Also, have two
cold spares in the rack. DL 320s are cheap and "rarely fail" using
1.2.X. I actually cannot remember a single failure over years of
operation.
I have no idea what you mean "if the primary fails". Are we using
Windows NT, does the secondary domain controller take over?
| Quote: |
The point is that most people that want a DS3 interface really do want
to pump in a DS3's worth of calls, more or less, in which case they
really can't afford to have those DS1s going spare just for redundancy's
sake. And if you are doing substantially less than a DS3's worth of
calls, you probably shouldn't be looking at a DS3 interface to begin
with unless that's just an incredibly lucrative way to get channelised
PRIs in from a vendor - and with typical the cost of UNE DS3 loops vs
T1s, that's not necessarily so.
|
I don't think you have much experience with DS3s, correct me if I am wrong.
While pricing many solutions, it is either 14 or 16 T1s where a DS3
becomes about the same cost for the loop, that is a lot of wiggle
room. Would you pay for 16 of something you need or take 28 of
something you surely will need down the road.
| Quote: |
Secondly, an industrial-grade ISDN media gateway designed for telco
environments (like a Cisco AS, say) isn't going to go down frequently
enough to merit this kind of concern. Don't get me wrong, I am the last
to go on record saying that Cisco voice equipment (or any other) doesn't
fail from time to time -- ha. But, again, 80/20 rule. A PC is much
more likely to fail within the same MTBF.
So yes, a single gateway handling a DS3 can go down. But so can an M13
mux. You've got single points of failure either way.
|
Read the specs on the Adtran 2800 MX13. I don't think it is going to
fail unless you smash it or pour coffee on it. Google it and RTFM
before you spout off about a product you obviously have no knowledge
of.
| Quote: |
If one is in the sort of environment where such high availability really
is a concern (typically a telco setting), one probably needs to invest
in a big DACS and redundant, protection-switched DS3 paths (and
protection line cards for them on the DACS side) as well as redundant
gateways, or at least redundant DS3 line cards in the chassis. At that
point of stringent availability, this discussion becomes a wee bit moot
because most likely you would not be using Asterisk and PCs in such a
setting anyway.
|
We are talking DS3 here, not OC12. Talk about overkill.
| Quote: |
| Quote: | Running stripped down Linux OS boxen with quad port T1 cards and four
or five lines in extensions.conf, no unneeded modules or software
loaded, asterisk 1.2, entries for zaptel and zapata, and a couple
entries in sip.conf builds a worry free solution. Asterisk uptime 2
years, system uptime two years and twenty minutes.
|
Yes, but total cost of ownership goes up because you need someone to do
all that, and even so, despite the impressive uptime you mention, PCs do
need a lot more maintenance, upkeep and worry.
|
Original setup takes a little time, reviewing the logs a couple of
times a weeks takes a minimal amount of time. Rotating logs should be
automatic, what else is there?
| Quote: |
With dedicated media gateways, you just plug in, set up and it works.
|
They are essentially small servers, some with solid state and flash,
others with real hard drives. Ever open one up? I would prefer to
pop a case and replace a T Card, memory, hardrive, powersupply, fan,
then waiting on an RMA. Especially when your call center is losing
$26k an hour.
Thanks,
Steve
PS, all I know is what works well for a 15k-20k calls a day center
with the average call lasting fifteen minutes and EVERY second of
every call being recorded. Maybe I am a newb. |
|
| Back to top |
|
|
benny+usenet at amorse...
Guest
|
| Posted: Mon Apr 07, 2008 7:05 am Post subject: [asterisk-users] Where is the Digium DS3 card? |
|
|
Tilghman Lesher <tilghman at mail.jeffandtilghman.com> writes:
| Quote: | And the arguments on the other side come down to "I'm using an ISP
which can't correctly configure their mailserver, and I'm too lazy to set one
up myself."
|
How can the mail server fix a broken reply-to? It can remove it of
course, but that is rather silly.
| Quote: | and "I'm too lazy to check the headers when I send out a reply."
|
Absolutely, I am. At least Gnus has a "broken-reply-to" setting that I
can toggle. It doesn't solve the problem that proper needed reply-to's
are removed too (it can't, since the mail server removed all traces of
them), but fortunately reply-to is almost unnecessary these days.
Maybe I should just set "broken-reply-to" for all groups, even the
correctly working ones.
/Benny |
|
| Back to top |
|
|
abalashov at evaristes...
Guest
|
| Posted: Mon Apr 07, 2008 9:46 am Post subject: [asterisk-users] Where is the Digium DS3 card? |
|
|
Steve Totaro wrote:
| Quote: | 7 HP DL320s, RAID 1 with Quad Sangoma. Not a dozen but more than
twice that, 28 T1s. What is your cheaper solution? Also, have two
cold spares in the rack. DL 320s are cheap and "rarely fail" using
1.2.X. I actually cannot remember a single failure over years of
operation.
|
My solution, if we may call it that, is certainly not cheaper. It is
not always about price. There's that whole value thing, too. 
The point I was trying to get at earlier is that if one has unused DS1
carriers on a DS3 and you are debating extra cold-spare PCs with T1
cards to provide failover in a PRI hunt group because one literally
cannot afford downtime at the statistical incidence typically incurred
by the failure of industrial DS3 interface equipment, one is probably
chasing their tail and needs to rethink one's strategy at a much higher
level than simply the question of terminating equipment.
| Quote: | I don't think you have much experience with DS3s, correct me if I am wrong.
|
I do.
| Quote: | While pricing many solutions, it is either 14 or 16 T1s where a DS3
becomes about the same cost for the loop, that is a lot of wiggle
room.
|
A bit of a digression, but:
That is commonly the case, indeed, with "many solutions." It is also
commonly not the case. Sadly, this topic lends itself to generalisation
not; if it did, our lives would be considerably simplified and a
marketplace rationale value chain consisting of a whole slew of agents,
resellers and brokers would cease to exist.
It all depends on who the transport vendor is, if it's an ILEC, and if
so, which ILEC, which LATA, applicable tariffs, forebearances, whether
there is inter-CO mileage, etc. If it's a CLEC, it's a question of
whether there are blacklisted COs in the span design, the price of their
CFA. If it's a retailer of Bell services it's all a question of what
kind of pricing and options _they_ get, and whether the DS3 is delivered
over regulated, UNE-subject copper, etc.
I have been in situations where it is more economical to get a DS3 after
about 5 T1s because the DS3 was built over leased utility dark fibre in
a metro area. I have also been in situations where even 20 T1s are
cheaper than a DS3 loop because of the particular facilities and various
pricing/tariff absurdities that are so characteristic of ILECs. I have
also seen on more than one occasion where a stupidly large number of
distinct analog lines (as in, pairs all the way from the MDF, not even
GR.303/DLC-delivered stuff!) was cheaper than using T1 as a carrier for
an equivalent number. Something about USF subsidies in rural areas
*cough* *cough*.
It all depends.
| Quote: | Read the specs on the Adtran 2800 MX13. I don't think it is going to
fail unless you smash it or pour coffee on it. Google it and RTFM
before you spout off about a product you obviously have no knowledge
of.
|
I have used the Adtran M13 muxes, and I postulate nothing about them
save that it has the common liabilities of all electronic devices
composed of matter and compliant with the laws of physics and the
discovered properties of the known natural universe. They can and will
break.
No, seriously, I agree -- be it an Adtran or a Widebank or whatever,
it's pretty hard for an M13 mux to blow. It's almost something you have
to really want to happen.
But do realise the sort of equipment toward which I am gesturing is
engineered to a similar level of reliability, proportionally, as much as
is possible in light of increased complexity and composition.
| Quote: | We are talking DS3 here, not OC12. Talk about overkill.
|
What is fundamentally different? Both are used as transport facility
for high-density inter-machine trunking. One does not need native SONET
to have a need for extremely high TDM circuit availability.
No, you do not need the level of redundancy I am suggesting in the sort
applications being discussed. That's the whole point. If, on the other
hand, you do, to the point where you're fretting about a DS3 interface
on a commercial VoIP gateway failing vs. having a bunch of PCs with
broken-out T1s, then it might be worth the investment to overengineer in
a manner that corresponds to telco environment requirements.
| Quote: | They are essentially small servers, some with solid state and flash,
others with real hard drives. Ever open one up? I would prefer to
pop a case and replace a T Card, memory, hardrive, powersupply, fan,
then waiting on an RMA. Especially when your call center is losing
$26k an hour.
|
This is true, PCs are easier to deal with when they fail. No question
about that.
On the other hand, Cisco AS equipment is that much less likely to fail.
Keep one cold spare around and you're good.
--
Alex Balashov
Evariste Systems
Web : http://www.evaristesys.com/
Tel : (+1) (678) 954-0670
Direct : (+1) (678) 954-0671
Mobile : (+1) (706) 338-8599 |
|
| Back to top |
|
|
jra at baylink.com
Guest
|
| Posted: Mon Apr 07, 2008 10:35 am Post subject: [asterisk-users] Where is the Digium DS3 card? |
|
|
On Mon, Apr 07, 2008 at 10:48:34AM +0100, Horwich IT Services wrote:
| Quote: | On Sun, 6 Apr 2008 17:22:58 -0400, "Jay R. Ashworth" <jra at baylink.com>
wrote:
| Quote: | Yes, I've seen that, and most of its arguments are specious, at best.
They amount to "I am too stupid to find a mail user agent with List
Reply, and too lazy to switch to it".
|
Are there any MUAs (other than Microsoft's pitiful offerings) that do not
observe RFC2369 headers?
|
The only one of those headers that seems germane to this thread is
List-Post... and I hadn't thought Mutt actually supported it, but on
looking, I find that it does.
Question is: does Mailman *set* it?
Cheers,
-- jra
--
Jay R. Ashworth Baylink jra at baylink.com
Designer The Things I Think RFC 2100
Ashworth & Associates http://baylink.pitas.com '87 e24
St Petersburg FL USA http://photo.imageinc.us +1 727 647 1274
Those who cast the vote decide nothing.
Those who count the vote decide everything.
-- (Joseph Stalin) |
|
| Back to top |
|
|
jra at baylink.com
Guest
|
| Posted: Mon Apr 07, 2008 10:39 am Post subject: [asterisk-users] Where is the Digium DS3 card? |
|
|
On Mon, Apr 07, 2008 at 02:46:48AM -0400, Alex Balashov wrote:
| Quote: | The point is that most people that want a DS3 interface really do want
to pump in a DS3's worth of calls, more or less, in which case they
really can't afford to have those DS1s going spare just for redundancy's
sake. And if you are doing substantially less than a DS3's worth of
calls, you probably shouldn't be looking at a DS3 interface to begin
with unless that's just an incredibly lucrative way to get channelised
PRIs in from a vendor - and with typical the cost of UNE DS3 loops vs
T1s, that's not necessarily so.
|
My understanding was that the breakeven point for T-3 transport was
something like 8-10 T-1's; no?
| Quote: | So yes, a single gateway handling a DS3 can go down. But so can an M13
mux. You've got single points of failure either way.
|
Sure. But are the gateways as redundant as, say, that Adtran mux that
Steve was raving about?
| Quote: | Yes, but total cost of ownership goes up because you need someone to do
all that, and even so, despite the impressive uptime you mention, PCs do
need a lot more maintenance, upkeep and worry.
With dedicated media gateways, you just plug in, set up and it works.
|
And that's a point.
Cheers,
-- jra
--
Jay R. Ashworth Baylink jra at baylink.com
Designer The Things I Think RFC 2100
Ashworth & Associates http://baylink.pitas.com '87 e24
St Petersburg FL USA http://photo.imageinc.us +1 727 647 1274
Those who cast the vote decide nothing.
Those who count the vote decide everything.
-- (Joseph Stalin) |
|
| Back to top |
|
|
g.stewart at horwits.c...
Guest
|
| Posted: Mon Apr 07, 2008 11:02 am Post subject: [asterisk-users] Where is the Digium DS3 card? |
|
|
On Mon, 7 Apr 2008 11:35:43 -0400, "Jay R. Ashworth" <jra at baylink.com>
wrote:
| Quote: | Question is: does Mailman *set* it?
|
Yes.
--
Godwin Stewart - Horwich IT services |
|
| Back to top |
|
|
tilghman at mail.jeffa...
Guest
|
| Posted: Mon Apr 07, 2008 11:11 am Post subject: [asterisk-users] Where is the Digium DS3 card? |
|
|
On Monday 07 April 2008 07:05, Benny Amorsen wrote:
| Quote: | Tilghman Lesher <tilghman at mail.jeffandtilghman.com> writes:
| Quote: | And the arguments on the other side come down to "I'm using an ISP
which can't correctly configure their mailserver, and I'm too lazy to set
one up myself."
|
How can the mail server fix a broken reply-to? It can remove it of
course, but that is rather silly.
|
You haven't read the "Reply-To Considered Harmful" article. The argument
is that some mail servers NEED the Reply-To set, because they (incorrectly)
send out mail with a From address that cannot be sent to. Correctly
configured mailservers don't need this hack, because they send out mail
with a proper From address.
| Quote: | | Quote: | and "I'm too lazy to check the headers when I send out a reply."
|
Absolutely, I am. At least Gnus has a "broken-reply-to" setting that I
can toggle. It doesn't solve the problem that proper needed reply-to's
are removed too (it can't, since the mail server removed all traces of
them), but fortunately reply-to is almost unnecessary these days.
Maybe I should just set "broken-reply-to" for all groups, even the
correctly working ones.
|
So the question comes down to, do you reply to the list more often or
do you reply off the list more often? Because the more frequent case
wins. In my case, I reply to the list more often, which I also believe to be
the case for MOST people; it is the EXTREMELY rare case that I ever reply
off-list. Hence, we make that case easier for the lazy, which means setting
the Reply-To in the listserv software. As a side effect, the list traffic is
also far more lively than it would be if we set the software NOT to set the
Reply-To header (which is a good thing IMHO).
--
Tilghman |
|
| Back to top |
|
|
aklists at mixdown.ca
Guest
|
| Posted: Mon Apr 07, 2008 12:08 pm Post subject: [asterisk-users] Where is the Digium DS3 card? |
|
|
On April 7, 2008 02:01:08 am Alex Balashov wrote:
| Quote: | A Lucent TNT Max outfitted with _plethoric_ VFCs might work okay. Apex
too, perhaps. Haven't tried to see how much it can handle when TDM->RTP
translation is required.
|
I'm curious; are the cpu/tdm/dsp requirements for 672 g729 rtp streams that
much higher than 672 v92 data streams? I have done work for a dialup ISP
that has probably 20 of the damn things running for quite some time now with
zero issues, and I can't imagine that the RTP requirements are higher than
v92's.
-A. |
|
| Back to top |
|
|
abalashov at evaristes...
Guest
|
| Posted: Mon Apr 07, 2008 12:30 pm Post subject: [asterisk-users] Where is the Digium DS3 card? |
|
|
Andrew Kohlsmith (lists) wrote:
| Quote: | On April 7, 2008 02:01:08 am Alex Balashov wrote:
| Quote: | A Lucent TNT Max outfitted with _plethoric_ VFCs might work okay. Apex
too, perhaps. Haven't tried to see how much it can handle when TDM->RTP
translation is required.
|
I'm curious; are the cpu/tdm/dsp requirements for 672 g729 rtp streams that
much higher than 672 v92 data streams? I have done work for a dialup ISP
that has probably 20 of the damn things running for quite some time now with
zero issues, and I can't imagine that the RTP requirements are higher than
v92's.
|
This I do not know. Never had a chance to try.
I imagine you can do more with less intensive codecs, so most likely
more G.711u streams can be supported than G.729.
--
Alex Balashov
Evariste Systems
Web : http://www.evaristesys.com/
Tel : (+1) (678) 954-0670
Direct : (+1) (678) 954-0671
Mobile : (+1) (706) 338-8599 |
|
| Back to top |
|
|
stotaro at totarotechn...
Guest
|
| Posted: Mon Apr 07, 2008 12:56 pm Post subject: [asterisk-users] Where is the Digium DS3 card? |
|
|
On Mon, Apr 7, 2008 at 10:46 AM, Alex Balashov
<abalashov at evaristesys.com> wrote:
| Quote: | Steve Totaro wrote:
| Quote: | 7 HP DL320s, RAID 1 with Quad Sangoma. Not a dozen but more than
twice that, 28 T1s. What is your cheaper solution? Also, have two
cold spares in the rack. DL 320s are cheap and "rarely fail" using
1.2.X. I actually cannot remember a single failure over years of
operation.
|
My solution, if we may call it that, is certainly not cheaper. It is
not always about price. There's that whole value thing, too.
The point I was trying to get at earlier is that if one has unused DS1
carriers on a DS3 and you are debating extra cold-spare PCs with T1
cards to provide failover in a PRI hunt group because one literally
cannot afford downtime at the statistical incidence typically incurred
by the failure of industrial DS3 interface equipment, one is probably
chasing their tail and needs to rethink one's strategy at a much higher
level than simply the question of terminating equipment.
|
Nobody said anything about unused DS1s except you. Please re-read the thread.
Cold spares are good to have around, how can you not argue that fact.
Rather than take a box out of a rack and spend however long it is
going to troubleshoot it, you can pop the cold spare in and be up and
running in the amount of time it takes Linux to boot and change the IP
and a couple of other settings, then return to the dead box. If it is
that level of redundancy is good enough for Computer Science
Corporation, it is good enough for me.
A cold Adtran MX2800 M13 spare is a good idea, isn't it?
| Quote: |
| Quote: | I don't think you have much experience with DS3s, correct me if I am wrong.
|
I do.
| Quote: | While pricing many solutions, it is either 14 or 16 T1s where a DS3
becomes about the same cost for the loop, that is a lot of wiggle
room.
|
A bit of a digression, but:
That is commonly the case, indeed, with "many solutions." It is also
commonly not the case. Sadly, this topic lends itself to generalisation
not; if it did, our lives would be considerably simplified and a
marketplace rationale value chain consisting of a whole slew of agents,
resellers and brokers would cease to exist.
It all depends on who the transport vendor is, if it's an ILEC, and if
so, which ILEC, which LATA, applicable tariffs, forebearances, whether
there is inter-CO mileage, etc. If it's a CLEC, it's a question of
whether there are blacklisted COs in the span design, the price of their
CFA. If it's a retailer of Bell services it's all a question of what
kind of pricing and options _they_ get, and whether the DS3 is delivered
over regulated, UNE-subject copper, etc.
I have been in situations where it is more economical to get a DS3 after
about 5 T1s because the DS3 was built over leased utility dark fibre in
a metro area. I have also been in situations where even 20 T1s are
cheaper than a DS3 loop because of the particular facilities and various
pricing/tariff absurdities that are so characteristic of ILECs. I have
also seen on more than one occasion where a stupidly large number of
distinct analog lines (as in, pairs all the way from the MDF, not even
GR.303/DLC-delivered stuff!) was cheaper than using T1 as a carrier for
an equivalent number. Something about USF subsidies in rural areas
*cough* *cough*.
It all depends.
|
On who you know.
| Quote: |
| Quote: | Read the specs on the Adtran 2800 MX13. I don't think it is going to
fail unless you smash it or pour coffee on it. Google it and RTFM
before you spout off about a product you obviously have no knowledge
of.
|
I have used the Adtran M13 muxes, and I postulate nothing about them
save that it has the common liabilities of all electronic devices
composed of matter and compliant with the laws of physics and the
discovered properties of the known natural universe. They can and will
break.
No, seriously, I agree -- be it an Adtran or a Widebank or whatever,
it's pretty hard for an M13 mux to blow. It's almost something you have
to really want to happen.
But do realise the sort of equipment toward which I am gesturing is
engineered to a similar level of reliability, proportionally, as much as
is possible in light of increased complexity and composition.
| Quote: | We are talking DS3 here, not OC12. Talk about overkill.
|
What is fundamentally different? Both are used as transport facility
for high-density inter-machine trunking. One does not need native SONET
to have a need for extremely high TDM circuit availability.
No, you do not need the level of redundancy I am suggesting in the sort
applications being discussed. That's the whole point. If, on the other
hand, you do, to the point where you're fretting about a DS3 interface
on a commercial VoIP gateway failing vs. having a bunch of PCs with
broken-out T1s, then it might be worth the investment to overengineer in
a manner that corresponds to telco environment requirements.
|
Yes and when your commercial grade VoIP gateway fails, you have
NOTHING. I am still up and running just to a lesser degree.
| Quote: |
| Quote: | They are essentially small servers, some with solid state and flash,
others with real hard drives. Ever open one up? I would prefer to
pop a case and replace a T Card, memory, hardrive, powersupply, fan,
then waiting on an RMA. Especially when your call center is losing
$26k an hour.
|
This is true, PCs are easier to deal with when they fail. No question
about that.
On the other hand, Cisco AS equipment is that much less likely to fail.
Keep one cold spare around and you're good.
|
Please provide citation of your "facts" And when it does fail, then
what? Total outage. Instead of losing $1k/hr you are losing $26k/hr
| Quote: |
--
Alex Balashov
Evariste Systems
Web : http://www.evaristesys.com/
Tel : (+1) (678) 954-0670
Direct : (+1) (678) 954-0671
Mobile : (+1) (706) 338-8599
|
Well all I can add is that I installed the system outlined above and
have had zero trouble with the TDM->SIP servers at all in over two
years. This is a multimillion dollar company and it works flawlessly.
BTW, thanks for taking down the the article you wrote by reverse
engineering my work and claiming you did it all and announcing it to
the list. That was really shady.
Thanks,
Steve Totaro |
|
| Back to top |
|
|
abalashov at evaristes...
Guest
|
| Posted: Mon Apr 07, 2008 5:20 pm Post subject: [asterisk-users] Where is the Digium DS3 card? |
|
|
Steve Totaro wrote:
| Quote: | Nobody said anything about unused DS1s except you. Please re-read the thread.
|
Perhaps I misunderstood?
What you appeared to imply in regards to the failover benefits of having
T1s broken out of an M13 mux into multiple gateway machines is that if
one of the gateways go out, calls can go roll over to other PRIs in a
trunk group. That suggests that there are unused PRIs with some sort of
substantial leftover capacity.
My point was that if you got enough capacity just sitting there to
really make this an economical enhancement from the standpoint of
managing statistical loss expectancy on a DS3 meaningfully, I am brought
to ask why to get a DS3 in the first place.
| Quote: | A cold Adtran MX2800 M13 spare is a good idea, isn't it?
|
No controversy there.
Most definitely, but far from always the sole determinant.
| Quote: | Yes and when your commercial grade VoIP gateway fails, you have
NOTHING. I am still up and running just to a lesser degree.
|
But that's why one keeps a cold spare around, right?
The axial thesis here is that if your uptime requirements are so tight
that you can't afford to be down for the time it takes to swap to a cold
spare gateway, and your chosen strategy to mitigate that is to use a
bunch of PCs with T1 cards, then there's a lot upside down here and
you've got far bigger fish to fry anyway. From a risk management
standpoint, not necessarily any other.
But that's the essence of the polemic about uptime, is it not?
| Quote: | | Quote: |
This is true, PCs are easier to deal with when they fail. No question
about that.
On the other hand, Cisco AS equipment is that much less likely to fail.
Keep one cold spare around and you're good.
|
Please provide citation of your "facts" And when it does fail, then
what? Total outage. Instead of losing $1k/hr you are losing $26k/hr
|
I don't have MTBF data on various equipment. Much as you do, I simply
furnish you with the narrative of my empirical experience; I've seen
lots of PCs die, lots of Digium/Sangoma/etc. cards die or perform
poorly, and I'm yet to see any Cisco voice chassis I administer die.
I've got some with 4+ year uptimes.
I'm sure it's coming soon.
| Quote: | Well all I can add is that I installed the system outlined above and
have had zero trouble with the TDM->SIP servers at all in over two
years. This is a multimillion dollar company and it works flawlessly.
|
Perhaps you are onto something, then. I would say you're extremely
fortunate whether you are or not. That has not been my experience, and
that is all I can attest to, really.
| Quote: |
BTW, thanks for taking down the the article you wrote by reverse
engineering my work and claiming you did it all and announcing it to
the list. That was really shady.
|
As mentioned before, I did not reverse engineer your work. I
documented the standard approach to the Asterisk/Hylafax problem, oft
repeated in many places, of which your implementation -- which I have
indeed seen -- was a good example.
Feel free to e-mail me off list any suggestions you have as to what
aspects of my article infringe upon any idiosyncratic or proprietary
aspects of your work or your implementation in particular, and I will be
more than delighted to work something out with you.
Thanks,
-- Alex
--
Alex Balashov
Evariste Systems
Web : http://www.evaristesys.com/
Tel : (+1) (678) 954-0670
Direct : (+1) (678) 954-0671
Mobile : (+1) (706) 338-8599 |
|
| Back to top |
|
|
|
|
|
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum
|
Powered by phpBB © 2001, 2005 phpBB Group
|
Search
