Archive for January, 2012

A Happy Holiday Season Indeed

Monday, January 16th, 2012

Having finished a hectic and rewarding December, we thought it would be worthwhile to go back and analyze activity on Christmas day and the week that followed, as it’s one of the busiest periods of the year for our digital clients.  We compared the data from this period to our “run-rate” business.  Hats off to my colleague, CTO Brett Thomas, for the numbers.

  • On December 25th, CashBox processed 3.77 million SOAP calls – think of a SOAP call as a “message” calling our system from any of our clients.  A SOAP call could be setting up an account, retrieving a customer record, initiating a refund, or any other activity.  Our typical daily run-rate for the month was approximately a million, which means we saw a Christmas day increase of 275%+
  • During the busiest period on that day, we were handling a peak load of 400 concurrent calls.  Most of these were related to account sign-ups and activations, as gift recipients went online in droves. This is about 10 times our normal peak load in 2011.
  • During the week between Christmas and New Year’s Day, we signed up nearly one million accounts for a client who experienced a very busy and successful digital shopping season.

As I’ve alluded to in recent posts, the Digital Economy has significant implications on scalability and uptime, and during very specific periods in the year.  The wave of activity can include Christmas week for digital retail; the lead up to Valentine’s Day for online dating sites; the start to each sports season, whether we’re talking the NBA, Nascar, NFL, or any other sports league; and the first day of a large scale MMO game launch.

Vindicia is rapidly closing in on one billion SOAP calls processed since the inception of CashBox, which we will hit in 2012.  Hitting this major milestone is something our whole team should be proud of; it highlights how we support and help grow the Digital Economy.  And this looks like only the beginning.

The Power of Nine

Monday, January 9th, 2012

My blog post last month on scalability generated many interesting internal and external conversations around reliability and uptime.  I am guilty of occasionally flippantly stating that Vindicia has a SLA that focuses on 99.99% uptime, but what this means in real life compared to, say, a guarantee of 99.9% uptime reveals fascinating business implications.

Here are the differences between the two uptime SLAs at a monthly level, courtesy Wikipedia:

99.9% = 43.2 minutes downtime

99.99% = 4.32 minutes downtime

The difference of nearly 39 minutes a month might not seem like a big deal.  However, if you are an online merchant that is going through a huge Christmas surge, that hypothetical downtime would create a significant business loss.  For example, if one of our clients would have experienced the additional downtime during their heaviest signup period in 2011, they would have lost over 50,000 new customers in that 39 minute span.  This is especially relevant because it is more likely that there would be a downtime during an extended peak surge.

If each customer is worth $20 that first year, that’s a million dollar mistake.  The reality is that the amount is greater because you have to account for the the lifetime value of those customers.  Also, don’t forget the opportunity cost:  all the customers you never sign in the first place because of the bad publicity.

Reliable infrastructure matters:  putting uptime numbers in the context of what clients do in the Digital Economy is incredibly revealing.