「予想外」のトラフィックに耐えられるか?

ソフトバンクの新料金プランは、“予想外”に複雑
ということなので、この話題はいったんお休み。

で、
NTT西日本、ひかり電話の通話状況は現時点で安定。引き続き監視を強化
となりましたが、NTT東でも同様の事故を思い起こし
ひかり電話はなぜトラブルが続くのか
という記事を読んでいて、やっぱり
「NTTがこれから構築を進めるNGN（次世代ネットワーク）になれば，こういった問題は起きないのだろうか。」
という疑問ですよね。

さて、ケータイの世界でも同じ問題があります。
一部のメディアで「ソフトバンクが定額制を出せたのはシェアが少なくてユーザ数がまだまだだから。同じことをドコモがしたら直ぐにっ戸ワークがダウンするよ」と書かれていました。まあどうなるかはこれからとして、こんな記事は要チェックでしょうね。

+++++
The message is the medium
By Tim Kridel
Oct. 26, 2006

To say that the fifth season of American Idol was popular with Cingular Wireless customers might be the biggest understatement since "Houston, we have a problem." The company handled more than 64.5 million text messages during the season--more than one text message for each of the 55.8 million customers it had at the time.

It's easy to assume that most wireless carriers dream about a promotion that successful, and that's correct--to an extent. But accommodating that much traffic also can be a nightmare, particularly in terms of making sure that infrastructure such as short message service centers (SMSCs) can handle sudden spikes, such as when millions of viewers all vote at once.

"It was certainly an engineering challenge," Megan Klenzak, director of technology at Cingular Wireless. "For the first few seasons, [there were] no forecasts and no firm expectations as to what peaks to expect."

That type of wing-and-a-prayer approach is a risky move for at least two reasons. First, if the network becomes overloaded to the point that messages are delivered late or not at all, then the contest might not be considered fair because not every voter and contestant had an equal chance. Second, regardless of whether they've a vote for William Hung or a note between friends, wireless messages are both popular and profitable, so chronic quality-of-service (QoS) problems can start butchering a cash cow.

"SMS and MMS are profitable services for most wireless carriers in North America," said Julien Blin, research analyst for wireless and mobile communications at IDC. "The carrier's average profit margin for SMS is around 80%."

But simply over-provisioning SMSCs isn't a viable or cost-effective way to keep up with sudden spikes and long-term growth. Part of the challenge has to do with SMS itself, which originally was designed for tasks such as distributing voice mail notifications. Today, however, it's used more like instant messaging (IM).

"If you consider the rates and performance of an IM server versus the rates of an SMSC, it's unbelievably different," said Neil Ostroff, product manager at Tekelec. "An IM server will do tens of thousands of messages per second. To imagine an SMSC doing tens of thousands of messages per second wasn't even comprehensible at the time SMS was defined."

SMS is based on a store-and-forward architecture. That worked fine in the first decade or so of wireless, when short battery life and spotty coverage meant that initial delivery attempts often failed. Then a few things happened: Batteries and coverage improved, rates came down, and IM on PCs became popular. As a result, wireless users began expecting SMS to be instantaneous.

One way to achieve that experience is by rethinking SMSCs. Tekelec's proposed approach is a distributed-router architecture.

"The future is that you want to have a distributed architecture of routers that are centrally managed, and that creates your SMSC," Ostroff said. "Now you have one big SMSC. All vendors will either move that way or they'll become obsolete."

Besides faster, IM-like service, the distributed approach also gives carriers more insight into messaging performance and thus greater control over QoS. Even so, most won't be willing simply to scrap their messaging infrastructure.

"This transition will not be an instantaneous one, so the traditional SMS-based messaging systems will still have their place within the carrier environment for some time to come," said Chris Melus, senior manager for data messaging development at Sprint. "This will be an evolution, not a revolution."

So to make the sale, vendors have to respect the past. In the case of Tekelec, that means using geographically distributed TekMedia routers to shoulder much of the load, including traffic spikes, message-waiting indicators (MWIs) and SIM updates.

"In many networks, the MWI is 50% of the mobile-terminated traffic," Ostroff said. "It's huge, especially in North America, where voice mail is free. The legacy SMSCs now will see only 10% to 20% of the traffic."

Vendor and carrier interest in alternatives also is driven by the slow migration to IP multimedia subsystem (IMS), including IP-based messaging, which enable a wider range of message types and adjunct services.

"The two that jump to mind are presence information and preference data," Sprint's Melus said. "User experiences will be greatly enhanced by having information as to what the recipient is doing and if they want to communicate."

For example, by looking at the recipient's preferences and device capabilities, the IMS infrastructure also will know whether to route messages to her PC or cell phone, and whether a message needs to be reformatted to accommodate that device's capabilities. In the process, the lines dividing services such as SMS, IM and e-mail will disappear, at least in the eyes of wireless users.

"As handsets and wireless networks start incorporating SIP and IMS, we should see the emergence of converged messaging services," said IDC's Blin. "IMS will change the way users messages each other. This will make the user unaware of the underlying technology, [such as] MMS, SMS, IM and e-mail. Users will have a single unified mailbox."

For wireless carriers, another benefit of IP-based messaging is greater flexibility, including the ability to develop market-differentiating services. That also reduces the need to compete primarily on price.

"In today's carrier environment, service and feature enhancements must always stay bounded by technical limitations of the technologies that enable them," Melus said. "In an all IP-based environment, everything is fluid and can be changed quickly and with minimal cost. This will enable carriers to try out new services with a small segment of their customers at minimal cost, time and exposure to risk. If the product or feature is well received it can quickly become available to the entire customer base without any loss of momentum while also incorporating changes from customer feedback."

But despite all of the new services that IMS enables--such as video IM and voicemail to MMS--the new infrastructure can't afford to ignore SMS and its still-growing customer base.

"The most important service of IMS messaging is support for SMS," said Tekelec's Ostroff. "It must have backward-compatibility with SMS. Otherwise, you'll launch a service that's worse than 2G messaging in terms of the user experience: The only community you'll have to send cool messages to is the early adopters."
+++++

昨年膨大なトラフィックをCingular Wirelessにもたらしたアメリカン・アイドルの話からケータイキャリアのあり方に至るTelephonyに掲載されていたストーリーです。まあ、一部、宣伝っぽいところもありますけどちょっと考えさせられますね。