DataCenterNews Asia Pacific - Specialist news for cloud & data center decision-makers
Story image

Exclusive Interview: How Sotic executed its cloud migration strategy

Fri, 26th Apr 2019
FYI, this story is more than a year old

Leo Mindel, founder of Sotic Ltd, is a speaker at the Cloud Migration Summit 2019 on May 14th in London.

TechDay chatted to him about the “need to knows” of moving to the cloud – including budgeting for ROI, disaster recovery and facilitating his award-winning company that provides professional websites/web-based solutions to some of the most high profile sports clients (including Six Nations, British - Irish Lions, World Sailing, golf tournaments, etc).

What were Sotic's top 3 drivers for moving to the cloud?

As the digital landscape continues to rapidly evolve, it is crucial that we position ourselves to not only serve our clients but also serve their clients. Being able to dynamically change and deploy new services replacing elements that become outdated is our number one driver.

The second consideration was capacity. On a typical match day, some of our clients' online presence can receive as much as 50,000 page visits per minute. The ability for cloud solutions to burst with the peaks and troughs was a key requirement. The cloud also provides several cost benefits in terms of maintenance and security.

What cloud approach do you employ at Sotic? How did you assess your operational readiness when planning the migration?

We've gone all-in with AWS; to a certain extent we see multi-cloud and hybrid-cloud as requiring compromises to retain a certain level of provider agnosticism and we instead wanted to take advantage of Cloud with maximum efficiency.  That said our core stack is driven by Gitlab and Docker, so it would be fairly easy to port that if we needed to.

What was the biggest challenge you faced when migrating?

We made the decision to move to the cloud in 2015 and by 2016 we had moved to WordPress CMS to support our migration plans.  Prior to that, we had our on-prem infrastructure in Maidenhead and we were using an enterprise CMS. Our main consideration was managing interactions between the old and the new especially when it came to legacy codes and client platforms.

To put this in context, we had a number of product-based eCommerce sites, main websites, event booking, statistical management systems and the company administrative system all on-prem. There was a lot to consider and one of the key considerations was when we could take sites ‘offline' for testing. With assets such as eCommerce and event booking, there was never a ‘perfect' time.

The final switch came in the spring of 2018.  At this point, we were running on a hybrid infrastructure and still about a year out from migrating fully to the cloud but we experienced an outage during a period of scheduled maintenance and were unable to bring our on-prem systems back to full service. We had to stick or twist and ultimately chose to focus our resource in moving over the remaining systems.

Within 12 hours we had all of our core systems operating on cloud and then gradually moved our secondary and tertiary assets over time. It was a challenging few months of working round-the-clock but I have a brilliant team and they rose to the challenge. The choice was the right one, we are 100% cloud-based and we've never looked back which is why I plan to share this at the Cloud Migration Summit.

Gartner also forecasted that by the end of 2020, 80% of organisations will overshoot their cloud Iaas budget due to a lack of cost optimization approaches. Why is it tricky to predict cloud spend?

Those are two different questions. Taking the latter first, personally, I think the primary uncertainty in predicting cloud spend is that of predicting your demand. The better you understand the characteristics of your business and thus service demand, the better you can predict your cloud spend.  That, of course, assumes you understand your cloud cost model;  and if you don't, you should!

Cloud makes it easy to throw resources at fixing a problem and of course spiralling costs. In terms of cost optimisation, read on…

What are Sotic doing to avoid blowing their cloud budget and how are you maximising the value of your chosen cloud architecture?

Just like optimising other aspects of your technology estate, we employ the use of predictive analysis.  We find that cloud cost optimisation very much rewards the time you invest in it and how well you pay attention to your usage data.

As well as having a few budget alarms setup we review our spend monthly, looking at variance and categories of highest cost. We try to purchase optimally, using a core of reserved instances for critical permanent resources together with a blend of on-demand and spot for more dynamic requirements.

We also pay attention to the Trusted Advisor Cost Optimisation dashboard and the Cost Optimisation well-architected pillar. We also keep an eye on new announcements from AWS, such as when they recently announced the cheaper AMD powered server instances.

As you can see it is a continuous process which, as I mentioned, takes time.

Do you plan on moving all applications to the cloud or will you maintain some on-premises and why?

The plan has always been to migrate all our assets to the cloud.  However, we had mapped out the migration strategy to cover a couple of years which would then allow us to continue testing.  Once we experimented and made the switch we never looked back. As at today, we've already moved pretty much everything to the cloud;  all-in as I said and we are very proud of it. All that's left on-prem is a minimal local back office function including file/print services.

Has the migration been successful? Can you give us any stats illustrating the benefits of your move to the cloud?

Yes, it has been a huge success and has changed our approach to development and delivery, but not without some snags as discussed earlier.

To give an example of stats, during the Guinness Six Nations Championships where we run three of the official sites, Sotic were able to scale up our client's site by more than a factor of 10, to meet huge spikes in traffic. Such a feat would have simply been impossible or cost-prohibitive to achieve without the cloud.

I think to truly maximise the power of the cloud, aside from the expected technical requirements, there's a need for outside-the-box thinking; and once you have a grasp on that, cloud migration is very achievable.

Meet Leo Mindel at the Cloud Migration Summit on May 14th in London, where he will be discussing “Complete Cloud: Delivering Business Objectives In A Fully Cloud-Based Environment”.

Follow us on: