The success stories from the public cloud do not stop. Constantly new web and mobile applications appear that rely on infrastructure-as-a-service (IaaS) offerings. This is just one reason that animates market researcher like IDC ($107 billion) and Gartner ($244 billion) to certify a supposedly golden future for the public cloud by the year 2017. The success of the public cloud cannot be argued away, the facts speak for themselves. However, there are currently mostly startups, and many times non-critical workloads moving to the public cloud. Talking to IT decision-makers from companies that have mature IT infrastructures, the reality looks quite different. This also shows our study on the European cloud market. In particular, the self-service and the complexity is underestimated by the public cloud providers, when they go to corporate customers. But also for young companies the public cloud is not necessarily the best choice. I recently spoke with Florian Gschwandtner (CEO) and Rene Giretzlehner (CTO) of Runtastic who decided against the public cloud and choose a business cloud, more precisely T-Systems.
Runtastic: From 0 to over 50 million downloads in three years
Runtastic consists of a series of apps for endurance, strength & toning, health & wellness and fitness, helping users to achieve their health and fitness goals.
The company was founded in 2009 in Linz (Austria) and has, due to its huge growth in the international mobile health and fitness industry, now locations in Linz, Vienna and San Francisco. The growth figures are impressive. After 100,000 downloads in July 2010, there are now over 50 million downloads (as of October 2013). Of these, 10,000,000 downloads alone in Germany, Austria and Switzerland, which roughly means more than one download per second. Runtastic has 20 million registered users, up to 1 million daily active users and more than 6,000,000 monthly active users.
In addition to the mobile apps Runtastic is also offering its own hardware to use the apps and services more conveniently. Another interesting service is “Story Running”, which should lead to more variety and action while running.
Runtastic: Reasons for the business cloud
Considering this growth numbers and the number of growing apps and services Runtastic is an ideal candidate for a cloud infrastructure. Compared to similar offerings even for a public cloud. You would think. Runtastic began with its infrastructure on a classic web host – without a cloud model. In recent years, some experience with different providers were collected. Due to the continuing expansion, new underlying conditions, the growing number of customers and the need for high availability the reason was to outsource the infrastructure into the cloud.
Here, a public cloud was never an option for Runtastic because of the dedicated(!) location of the data as a top priority, not only because of security concerns or the NSA scandal. So, the Amazon Web Services (AWS) just for technical reasons are out of the question, since many demands of Runtastic can not be mapped. Among other things, the core database comprises about 500GB on a single MySQL node, which AWS can not meet, according to Giretzlehner. In addition, the cost for such large servers in a public cloud are so expensive that an own hardware quickly pays for itself.
Instead Runtastic decided for the business cloud of T-Systems. This was due to a good setup, a high quality standard and two data centers in one location (in two separate buildings). In addition, a high level of security, a professional technical service as well as the expertise of the staff and partners and a very good connection to the Internet hub in Vienna and Frankfurt.
To that end Runtastic combines the colocation model with a cloud approach. This means that Runtastic covers the base load with its own hardware and absorbs peak loads with infrastructure-as-a-service (IaaS) computing power of T-Systems. The management of the infrastructure is automated by Chef. Although the majority of the infrastructure is virtualized. However Runtastic drives a hybrid approach. This means that the core database, the storage and the NoSQL clusters are operated on bare metal (non-virtualized). Because of the global reach (90 T-Systems data centers worldwide) Runtastic can continue to ensure that all apps work seamlessly and everywhere.
A unconventional decision is the three-year contract, Runtastic closed down with T-Systems. These include the housing of the central infrastructure, a redundant Internet connection and the demand of computing power and storage of the T-Systems cloud. Runtastic chose this model because they see the partnership with T-Systems in the long term and do not want to leave the infrastructure from one day to another.
Another important reason for a business cloud was the direct line to counterparts and professional services. This also reflects in a higher price, what Runtastic consciously takes into account.
The public cloud is not absolutely necessary
Runtastic is the best example that young companies can be successful without a public cloud and that a business cloud is definitely an option. Although Runtastic runs not completely on a cloud infrastructure. However, the business case must be calculated in each case, whether it is (technically) worthwhile to invest parts in an own infrastructure and intercept the peak by cloud infrastructure.
This example also shows, that the public cloud is not easy sledding for providers and consulting and professional services from a business or managed cloud are in demand.
During the keynote at AWS re:Invent 2013 Amazon AWS presented its flagship customer Netflix and the Netflix cloud platform (see the video from 11 minutes), which was developed specifically for the Amazon cloud infrastructure. Amazon sells the Netflix tools of course as a positive aspect, what they definitely are. However, Amazon conceals the effort that Netflix operates in order to use the Amazon Web Services.
Netflix shows very impressive how a public cloud infrastructure is used via self-service. However, when you consider what an effort Netflix operates to be successful in the cloud, you just have to say that cloud computing is not simple and a cloud infrastructure, no matter what kind of provider, needs to be built with the corresponding architecture. This means, conversely, that the use of the cloud does not always lead to the promised cost advantages. In addition to savings in infrastructure costs which are always pre-calculated, the other costs may never be neglected for the staff with the appropriate skills and the costs for the development of scalable and fault-tolerant application in the cloud.
In this context, the excessive use of infrastructure-near services should be reconsidered. Although these simplify the developers life. As a company, you should think in advance whether these services are actually absolutely needed to keep your options open. Virtual machines and standard workloads are relatively easy to move. For services that engage very close to the own application architecture, it looks different.