Curtis, re: the Bell South outages in your post, I find it very interesting, and corroborating to a certain degree, that the Bells have not assigned the same level of criticality and importance that they have, traditionally, to their bread and butter POTS services. One might argue that this is a philosophical issue.
While I don't agree with this (as I demonstrate in a moment), there are actually some plausible grounds for such an argument.
The argument might go something like this: Voice services are more important, more critical, because of the lifeline connotations usually assigned to them. Furthermore, the residential and small business DSL service might be viewed with less criticality due to the casual nature which is usually assigned to web surfing, especially if the DSL service is sold as "residential" grade. Finally, no one can guarantee end user performance, in any event, due to the vagaries associated with upstream conditions on the 'Net, making overall assessment of performance 'iffy,' at best.
I emphasize residential here, as opposed to one that has been sold with service guarantees, such as one that would be used for commerce. Covad, Northpoint, and a growing list of others come to mind, where SLAs and QoS-like guarantees come into play. The Bells also supply these. But no matter who supplies these higher grades of service, they generally run some five to ten times more expensive than those which are designated for "residential" use.
[And alas, from what I've been reading on NANOG and in other press stories, even these specialized firms which I began listing above, who cater exclusively to enterprises, themselves, have been having an horrific time keeping themselves above water, due, in part, to the inability of their host providers (the ILECs, usually) to keep their platforms up and working. Maybe that's not fair, because I don't know all of the specifics in these episodes, although the articles I've read point to the hosting entities. What is for certain, though, is that they've not been all that reliable in a surprisingly high number of situations. Here, I'm referring to outage scenarios similar to the one which your post highlights, where entire cities and regions have been going up and down for days on end.]]
It's clear that these outages have resulted from transitioning from an older set of provisions to a newer one. Or, in some manner there was planned implementation work taking place designed to upgrade the platforms that were being used.
As such, whether the service was deemed casual or critical, really shouldn't have mattered. Leaving users vulnerable to such outages without making adequate provisions for contingency situations is simply uncharacteristic of usual Telco practices (for voice) and should be regarded as totally unacceptable, by any measure. It surprises me further to read a spokesman of the company describing what has taken place, as though there was something akin to normalcy about it, and with relative impunity.
In contrast, anytime a new voice switch is introduced in a "working condition", i.e., where live services exist, the "cutover" or swapping out of the old with the new generally goes like clock work, because adequate provisions are always made for POTS.
POTS services, you see, are "measured services," which generally means that they are actively measured by the PUCs. There is a performance measurement, or "index" which is used to evaluate every conceivable parameter of POTS.
These indexes, in turn, become crucial yardsticks when it comes time for evaluating LEC requests for rate increases, and other favorable treatments (such as LD approvals, mergers, licensing for new services, etc). Witness, the quality of service issues now being examined in USW territory in the assessment of their merger approval with Q.
Precautions are usually invoked when swapping out voice provisions, such as through the use of dual stored configurations, which effectively allows the carrier to re-invoke the original configuration in a contingency, in other words, should something go wrong, so that the transition that took place at 3A.M. Sunday morning went unnoticed to the consumer. Some of these things are Network Implementation 101, as any enterprise engineer or ISP jock would tell you. You just don't leave customers swinging in the wind.
A key item here is that PUC metrics are collected for POTS services, measuring dial tone delay, percent of blocked calls, switch downtime, etc. I don't believe that such measurements are made for DSL just yet.
FAC |
