These days it is very rare to run out of trunks or switching capacity. However, during emergency situations, such as after heavy storms or other natural disasters, or other major emergencies, there may be a temporary overload on telephone facilities.
If this happens the switches may get overloaded. If a caller hits a busy signal (of any type), they'll try again very quickly. These unsuccessful call attempts represent a load on a switch.
Years ago a flood of call requests to an ESS might cause the entire switch to fail; I'm not sure if today's switches have the same limitation.
Some time ago the Bell System Technical Journal did a study on subscriber behavior out of concern that common control offices would be flooded in special situations. While somewhat dated, it is interesting.
My recollection is that, like everything else, the No.1 ESS was much more sophisticated at handling such issues than equipment studied in that article.
If there were an overload in a 1 or 1A ESS the ordinary subscriber would be denied dial tone. Certain priority subscribers could still be provided dial tone.
If a lot of subscribers remained off-hook, no harm, no foul, unlike a No
5 XBAR, which could crash given sufficient permanent off-hooks.
The No. 1 and 1A would eventually dump the permanent off-hook to the receiver off-hook (ROC) routine, then to suspend status ("dirty" battery). If traffic were really busy, the ROC tone would not be provided.
Nonetheless, the switch would be more apt to eventually provide dial tone to a subscriber who remained off-hook (that is, until ROC kicked-in) as opposed to a subscriber who would repeatedly plunge the switch-hook.
No doubt the No 5ESS would handlle the situation much the same because of similar design policies for such matters.
All of this presumes no major damage to the switch itself or its trunk frams. The loss of a line frame was insiginificant except for the subscribers assigned to that frame. (line module on a 5ESS I believe.)
the biggest traffic surges are literally man (or possibly teenager) made - in the UK the last one i heard about was due to tickets to a Pop concert going on sale....
and modems can be even worse.....
1 option in switches such as DMS 100 is "call gapping" which AFAICT limits the new connection rate sent across a trunk.
On a slightly different but somewhat related topic:
I seem to recall ILECs complaining that long modem connections were playing havoc with their provisioning plans... despite the growth of the overall market, the fate of companies like AOL suggests that the number of dialup users has shrunk. I wonder how much the factors used in provisioning planning changed as broadband replaced dialup as the most common connection type? Perhaps the high availability of trunk capacity mentioned in this thread is - at least in part - the result of capacity added during the dialup heyday and underutilized since? Or have technology upgrades over the past decade or two simply provided so much capacity that voice traffic can't normally put a dent in it?
I believe the heydey of dial-up was about 10-20 years ago, and at that time the local plant had older, more expensive equipment and subscriber loops. There was still quite a bit of analog ESS and even a few crossbar switches in service.
IMHO, the enormous drop in the cost of electronics allowed for far greater capacity. For instance, in 1975, the high cost of terminal equipment made it in some cases more economical to simply use plain copper as inter-office trunks as opposed to a higher capacity multiplexing scheme (per Bell System text). Plus, of course, people who have broadband pay for it as opposed to many dial-up users who merely used their regular home phone line.
I remember three occasions when dial tone or access was delayed. The first time was in Dallas where I was served by a step-by-step office (most of Dallas was step at that time.) It was during a storm and after I had tried two of three times to get dial tone I finally left the phone off-hook. About 40 minutes later I check again and I was getting dial tone, and when I dialed my numbeer it went right through.
The second time was in Kirkwood, Mo. (a suburb of St. Louis), again during a storm. It was a 5XB. You couldn't just leave your phone off-hook and wait because it would shortly give the off-hook tone. So I tried at intervals for several times, hoping perhaps to hit a time when there was a momentary gap in demand for service or the overload was eased. It took about 40 minutes that time, as I recall.
The third time was somewhat different, this time I was away from home a couple of years ago when a tornado alert was sounded and I wanted to call home to see if my wife was OK. For about an hour there was no signal.
I think that's it. Electronic switching and fiber trunking got rid of the bottlenecks in switching and transmission that constrained traditional telephony.
Not that the ILECs would ever admit that, of course.
Dial-up must be almost dead. And, there is a lot of excess capacity because of migtration away from wireline to wireless. Having said that, I suspect maintence of trunks as well as outside plant has declined. When there is excess trunk capacity it is cheaper to pickle a bad trunk rather than spend time and money repairing it. Same goes for local plant copper.
I suspect most of the LEC maintenance goes towards DSL plant these days.
On Thu, 05 May 2011 17:41:56 -0400, Geoffrey Welsh wrote: ........
If things are anything like in Australia, most intra-exchange backbone trunking is now VoIP so the "circuits" are being continually upgraded to handle the steadily increasing demands - whether voice or data.
The only issues I can see would be the number of trunk circuits coming out of the old voice switches and going into the VoIP carrier equipment.
I think there was also a large issue with reciprocal compensation. Most dial-up users were on ILECs, while the ISPs were on CLECs. The dial-up user often had flat rate local service, but the ILEC had to pay the CLEC per minute of connect time. I suspect a lot of money changed hands on this. I guess it's similar to today's compensation to conference call bridge companies, though I suspect dial-up ISPs resulted in much more traffic.
That's because the 5 XBAR used the markers for everything and being that markers were nothing but logical arragangements of relays, and limited depending on the office design, it was a fairly simple matter to overwhelm them.
But those markers were the second or third brush at common control. The hybrid then digital systems that replaced them benefited from lessons learned.
Well, a lot of the ILEC traffic is moving off the traditional switched network into VoIP services. Sure, it ultimately hits a switch somewhere in the chain but it's a purely digital signal so easy to mix into a time or frequency domain.
I'm not an expert on switching, but I do not think the above is correct. According to the Bell Labs history, the "marker" isn't called in until all the digits have been registered in the "originating register" and the "line link". So, I'm not sure the marker can be overwhelmed by numerous off-hooks.
I don't know if a crossbar switch could "crash" from being overwhelmed with traffic--"crash" being the system fails to work at all and needs external intervention to reset it (a "reboot"). Rather, I think call requests, either subscribers or external trunks coming in, would simply get slow service until traffic levels reduced.
The Bell System recognized the need for network control fairly early on and if too many calls were flooding a particular exchange or region they would manually busy out the trunks so the region wouldn't get overloaded. From panel onward switches contained numerous checks and alarms would sound if there were problems.
Also, both panel and crossbar had timers on off hook so that if the subscriber didn't complete their dialing within a specified time, the receiving equipment was cut off to avoid tying it up.
Other BSTJ articles discussed what happens to subscribers if there are too many off-hook (people trying to make calls) all at once. Obviously this was something they had to provide for very early on as emergency situations (eg major storms) would spark such calling. Later, radio promotions and the like could do so. One issue was whether a subscriber had "a place in line" while waiting, that is, were they served in turn, or randomly picked when a receiver became available. The BSTJ also gives rather complex statistical analysis formulas on all of this (as well of course trunk group usage and blocking).
A telephone switch isn't a mechanical device any more. Look at what your monitor is connected to and you'll see what a telephone switch is now. That's right, it's a computer, albeit a single-purpose one and a lot bigger. Note that nowadays you can run a PBX for a small to medium-sized office (and all a PBX is is a telephone switch that the customer owns instead of the telephone company) using Asterix on a dedicated PC.
The hardware reliability for phone company switches might be higher as the components aren't the least expensive ones, but even todays PCs are fairly high-reliability machines. The software running the switch, on the other hand, is much more reliable, it's equivalent to BSD or Linux uptime levels, e.g. 99.97% reliability.
I don't know the technical reasons but I do know that 65 permanent ROHs would crash a No. 5 XBAR. That was one of the reasons for 24/7 maintenance staff to pickel those ROHes, thus rendering them harmless.
I'm surprised ROH could do that kind of damage. People taking their phone off the hook to not get called, or, people leaving thier phone off accidently are both extremely common. In addition,a short circuit in a loop line may cause an off-hook. Is this ROH situation described in the literature anywhere?
According to the literature, ROH was forseen as an issue back in panel development, and timers were included to cut out the subscriber after a time interval has passed. This timer also applied while dialing. Timers continued in crossbar, and also included internally as call functions were pased from one unit to another--if the next unit didn't pick up in time, alternative action was taken. The switch had redundancy built into it.
Being a high complex electro-mechanical device often serving many thousands of subscribers there was a risk of failure. If a key component failed the switch could work around it but obviously traffic capacity would be reduced and dial tone might be delayed. Accordingly, a large switch may get 24/7 staffing, although some of that may be to take advantage of light traffic to do maintenance.
Certainly a high volume of traffic, either from subscribers or inter- exchange trunks, would slow processing. But the literature in BSTJ suggests the switch would not "crash"--that is totally shut down--but rather just be slow in handling calls. It might work less efficiently as components spend more time than normal trying to find alternate paths over busy components or busy trunks, but eventually a call would go through.
Perhaps I am splitting hairs on the difference between "no service at all" and "very slow service", but to me there is a big difference between a "system crash" where everything has failed and a restart is necessary, and a "systems slowdown" that is a temporary problem.
If the subscriber remains off-hook waiting for dial tone in a very busy condition I don't believe the ROH routine would come into play until after dial tone were provided and not acted upon in the prescribed time.
As to the No 1/1A dropping to dirty battery after completion of the ROH routine, the continuous popping sounds you then heard were sounds of orders (machine instructions) being issued by the processor.
Cabling-Design.com Forums website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.