IT is seen not just as a business enabler, but also as a division that can help the business identify new revenue...
streams. So when IT was becoming a limiting factor at Holiday Extras, a UK travel add-ons service provider, its IT team knew it had to act fast.
Holiday Extras provides add-on travel services, such as airport hotel bookings, parking and insurance services to British holidaymakers. Its most critical business application is its website.
The company’s in-house datacentre, which has been operational since its launch in 1983, powered its web applications, television campaigns and business operations. It was a regular facility full with blade servers, firewall infrastructure, load balancers and massive infrastructure around its web system.
However, Holiday Extras head of IT Damien Turner says in 2011 his team had noticed IT was consistently problematic.
While all the different divisions of the company had big, innovative and creative ideas to move the business forward, when it came to them asking for new hardware and software services to power those ideas, the IT team were rarely able to say "yes".
The tipping point was in September 2011 when the marketing team wanted to run a television campaign. Marketing told the IT team that to run a campaign during peak television shows, such as Coronation Street or Emmerdale, it would need to run 1,900 user sessions.
“So, the IT team did the load testing to see if our in-house IT kit could support the campaign," Turner explains.
"But we found that our infrastructure’s capacity was just 790 user sessions. After that, we would see red spots."
When the IT team assessed what it would take to fix this, it found it would have to increase the bandwidth, add new blade chassis and increase hardware infrastructure.
More on Amazon Web Services (AWS)
“We also found out it would take us between six and seven months to make our in-house IT kit ready if the entire team was to do nothing else but that project. That’s when we thought there must be an easier and better way to provision IT,” says Turner.
“It was around the same time that cloud computing was emerging as a feasible alternative to in-house infrastructure issues."
Because of the peaky nature of the holiday business, the company did not want to spend on hardware resources which will be capable of improving IT but will remain idle during seasonal lulls, making the capital investment go to waste.
Turner and his team picked Amazon Web Services (AWS) public cloud services.
“We were familiar with public cloud because we had just moved our email system to Google Apps. The team then simply lifted some of its in-house kit and hosted it on AWS platforms to benefit from cloud’s scalability, agility and flexibility.
“It was a bit strange because the technical team did not have a physical server to love and stroke, or turn on and off when something went wrong. The machine was virtual and just a name on the command line on the screen. But once it overcame the initial culture shock around cloud IT, it was smooth sailing," says Turner.
Before 2011, Holiday Extra’s capital expenditure was in the region of £1.6m.
“This was just the hardware kit – when we added the cost of the uninterruptible power supply (UPS); power; air-conditioners for the datacentre; and managing these additional bits, cloud computing seemed its best bet," he explains.
Using Amazon Web Services
Holiday Extras now uses AWS services, including Simple Storage Service (S3), Elastic Compute Cloud (EC2), CloudFormation, Elastic Load Balancing (ELB), Relational Database Service (RDS), MemCache and RedShift data-warehousing service.
The project involved moving 10 quad-core blade web servers – and two 1TB MySQL databases on the back end – to 12 web servers (across three availability zones), one large Amazon RDS instance, a small number of memcached instances, and 10 ELBs.
The cloud-based infrastructure has helped the company reduce IT hardware spend by more than half and cut user response times by 20%.
Investing in public cloud
But was it easy to convince the business to invest in public cloud when it was still in its infancy?
"2011 was when cloud was new and there was a perception that it is unreliable," says Turner.
My IT team is more of a DevOps team now. Focusing on strategic IT projects – such as coding for new business ideas rather than managing infrastructure – keeps the IT team more engaged and interested
Damien Turner, Holiday Extras
“Once we changed the perception of the cloud by helping everyone understand its benefits, it was easy to implement a cloud strategy. But our business stakeholders are visionaries and were keen to test the cloud services out."
To instill cloud confidence among Holiday Extras staff, the IT team started planning its strategies with cloud outages in mind.
“We would make provisions in our scripts to make them more tolerant of failures. We also have our internal datacentre operational to take over should there be an outage,” says Turner.
To ensure redundancy of its cloud infrastructure, Holiday Extras followed cloud computing best practice and has architected across three availability zones in AWS’ EU West region. This ensures high availability for its customer-facing website and allows the IT team to maintain the high-quality customer web experience, even when one availability zone experiences disruption.
The team experienced a four-hour downtime when the transition to the cloud happened. There were also initial teething problems – it would take the the team 45 minutes to start a server instance on the AWS infrastructure.
“This was because of the legacy codebase. It was more of a lump and we had to change the codebase to be able to provision servers more quickly. But even 45 minutes was a huge improvement on the six months that were required previously,” says Turner.
Enjoying the benefits of public cloud
Now, the cloud-savvy team is able to start up a server in seven minutes.
Other benefits of moving to the cloud include faster roll-out of new technologies, increased focus on customers, lower operating expenditure and flexible working.
But most importantly, IT has been able to respond to business changes quickly.
“We are spotting a new trend – our peak continues until September. I would have to worry about capacity planning if we were in our in-house infrastructure. Now, it is easy to scale the infrastructure up for as long as we need it.
"My IT team is more of a DevOps team now. Focusing on strategic IT projects – such as coding for new business ideas rather than managing infrastructure – keeps the IT team more engaged and interested," Turner explains.
However, Holiday Extras’ core booking system still runs on its in-house mainframe system.
“The core booking system was created 24 years ago. As the business evolved, we added more layers – such as payment applications and parking add-on applications – so it's not easy to just move it to the cloud. Now we are slowly trying to take each layer out and run it as a separate workload on AWS," he adds.
The mission-critical booking system is expected to move to the cloud in two years.
“We can now move faster and more cost-efficiently than our competitors, meaning more focus on retaining and gaining customers and less on the infrastructure. But above all, IT has become a "yes" department, thanks to AWS public cloud," Turner concludes.