By David Terrar, CEO of D2C Limited, Co-founder of Cloud Advocates
A few weeks ago I was one of the many worried parents and students trying to log in to the UCAS Track service on A Level results day to check the status of my daughter's University application. She hadn't quite got the required grades for her chosen course, but maybe she would be accepted by one of her two choices anyway? The only way to find out was the UCAS online system or by phoning the University in question. This year there were special circumstances to do with the increase in fees starting 2012 that meant there would be a lot more students chasing 2011 places. Whoever is in charge of the UCAS IT infrastructure didn't predict that hike in demand, and their system was down for most of the one day it was really needed - very poor planning in my opinion. Actually most of the Universities help lines were inundated and couldn't cope too. Putting aside the phone operators, this is EXACTLY the kind of capacity issue that Cloud Computing can help with.
Long ago in the 80s in another life one of my customers was the Cambridge examination board (now called Cambridge International Examinations). They had to buy the computing power required to manage the two peaks in their year when they had to process all of the marks and results just after exam time. For most of the rest of the year those processors (IBM System/38 for anyone who is interested) were only using a fraction of their capacity, but they absolutely had to perform at results time. That's normal for any IT operation (or website) - you have to cater for the potential swings in demand. That's where elastic Cloud Computing comes in and changes the old IT rules.
Probably the best known implementation of the idea is Amazon's Elastic Compute Cloud (EC2), but you can buy this kind of service from many suppliers. It combines with their standard infrastructure web service. Users can rent virtual machines to deploy their specific applications (or Software as a Service offering). Users can create, and terminate as many server instances as they need on demand, and pay by the hour for the required capacity - that's why it's called elastic. Set up correctly you can monitor demand, watch for the increased traffic and rent more short term capacity to manage a peak load. Why hadn't UCAS planned their IT infrastructure like that?
The best example of this that I've heard of recently is Comic Relief. Marcus East, their CIO, came to this year's Cloud Computing World Forum to tell the story. The lion's share of the charity's fund raising happens during the single event that we know and love - Red Nose Day. They have banks of phones and hundreds of volunteers on the day to accept telephone donations - the underlying IT systems that accept those payments HAVE to work. More than that, there may be a particular video shown on TV on the night which triggers many thousands of additional callers at one moment. The only practical and cost effective way of handling those massive peaks in demand is elastic computing. Marcus has worked with his partners at Carrenza, Cisco, Oracle, 3PAR, WorldPay, PayPal, VMware, Zeus Technology, and BT to make sure every donation gets processed. It works brilliantly. Marcus told me:
“Cloud computing is the only way that we can be certain of the scalability and flexibility needed to support the demands of our event night and we’re very grateful to our partners for their assistance.”
This is where Cloud Computing changes the game over the old approach. I just wish that the IT guys at UCAS were thinking this way.
David Terrar is a consultant and software developer who specialises in the use of Cloud applications and social media in business. He is a co founder of Cloud Advocates, an association of consultants who aim to demystify the Cloud and provide pragmatic help and advice for businesses, organizations and accounting practices. To find out more, visit cloudadvocates.com
Join us on