The cost of real time processing – the race with Moore’s Law

Written by on July 29, 2014 in BillingViews, News with 2 Comments

One of the things that we were not asked to do when writing the TM Forum report on real time charging and policy management was to look at cost. But we did anyway.

We spend a lot of time discussing (and now implementing) real time charging and policy management under the agenda item called ‘business models.’ But we tend to forget that Moore’s Law has been working away in the background and ten years have passed since we started talking about real time and its benefits.

We asked Shane O’Flynn, VP of Product Management with Openet about the cost angle and how much it has changed things. His response was that, “basically the capability to do the work in real-time has existed for decades it was simply too expensive to justify.” You had, in other words, to pick those processes that justified the cost.

And the cost was daunting. As O’Flynn says, “a little over ten years ago our largest customer was processing around three hundred million call records a day and we were using around $5M worth of hardware to do the heavy lifting. Doing a rough calculation, today we can process those same business rules and that volume of traffic on a single two CPU server at a cost of around $30,000.”

That makes the justification for the investment a whole lot easier. Essentially, today, anything that you want to process in real time, can be processed in real time without going white at the cost implications. Now, then, the “real-time versus batch [decision] is centred on the inherent value that you or your subscribers will derive from getting that information or that result in real-time.” 

Clearly, therefore, cash balances, advice of charge, accepting ‘spur of the moment’ offers all fall under the heading of ‘appropriate to process in real time.’ So, on one level, why would you not process everything in real time? “What value,” O’Flynn pondered, “do you derive by waiting?”

It would be easy to completely agree and say ‘none,’ but there are operators out there who believe that near real time is good enough. There are others out there – one MVNO as an example – who processes his post paid customers through two systems. First, he processes them through the prepaid system, which “addresses the real time needs of fraud control.” Then, at the end of the month, he runs them through the post paid system, “in order to make sure that all the bundles, discounts and offers have been accounted for, since you cannot do that in real time.”

We have this theory about progress on BillingViews Boulevard that says that an industry becomes obsessed by what it is capable of doing. For instance, when desktop publishing was invented, many, many companies started pushing out newsletters, because they could. Further, they used desktop publishing techniques to do every part of the newsletter, which was insane. You use the technology as long as it does the job better, faster or cheaper – and then you reach for the scissors (in the case of desktop publishing). The same was true of e-billing in all its different iterations and the same might be said to be true of real time charging and responsiveness.

Just because we now can do everything in real time, does not mean it is the most effective way of supporting our customers and our services. Use common sense – and then bring in the machines.

Tags: , , , ,

About the Author

About the Author: Alex was Founder and CEO of the Global Billing Association (GBA), a trade body focused on the communications sector. He is a sought after speaker and chairman at leading industry conferences, and is widely published in communications magazines around the world. Until it closed, he was Contributing Editor, OSS/BSS for Connected Planet. He is publisher of DisruptiveViews and previously BillingViews. .


If you enjoyed this article, subscribe now to receive more just like it.

Subscribe via RSS Feed

2 Reader Comments

Trackback URL Comments RSS Feed

  1. Bob Machin says:

    Seems to me there are three good arguments for real-time processing:
    > revenue assurance, particularly with respect to the credit-risky customer or partner
    > customer experience – being able to respond quickly and appropriately to certain customer behaviour, whether to protect them or to otherwise enhance their experience
    > to get through the work – if you’re having to process bazillions of transactions, there may be no other practical way of getting through them other than to process them immediately.
    ‘Because we can’ is still quite a good one, though…
    I believe there are two ‘n’s in O’Flynn, by the way…

  2. Alex Leslie says:

    Good points Bob, thank you. The RA angle needs some further investigation. And yes, two ‘n’s – thank you.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.