We are entering the second wave of cloud computing: The Public Cloud Economics.
Most enterprises understand cloud topologies (virtualisation) and privacy levels (private, virtual-private, and public), or simply the different resource types (IaaS, PaaS, or SaaS).
Some have even embraced pretty sophisticated technologies like cloud bursting – the dynamic relocation of workloads. However, compared to this sophisticated understanding of technology, understanding of the current or even future economic models of cloud computing lags behind.
More than a year ago, Forrester introduced the corporate perspective of cloud economics with James Staten’s report Drive Savings And Profits With Cloud Economics. The major cloud providers surprised us also with many innovative business models, such as Amazon’s AWS Reserved Instance Marketplace last September.
As an alternative to the fully flexible on-demand model, customers can also buy a one- or three-year contract for a compute instance and could save up to 60 percent. However, the risk is that you buy more than you need or simply the wrong instance type. The marketplace now allows organisations to sell off these half-used contracts to other customers.
The variety of different global cloud providers, local players, and even private cloud capacity, combined with different contract types such as the Amazon on-demand and reserved instance, stimulated the economic model of a cloud broker. See my reports about the cloud broker business model or the transformation process in larger enterprises.
While only very few large scale corporate IT divisions managed to establish a cloud broker style engagement model with their lines of business, some enterprises or government agencies use external cloud brokers. This external service meets the demand for dynamic sourcing options. The state of Texas, for example, is using the cloud broker Gravitant.com across many state agencies.
Despite these new cloud economics, the perspective was still limited by the corporate resources or the selected resources of a specific cloud broker.
This week’s announcement of the German Stock Exchange, Deutsche Börse, has accelerated cloud economics to the public level. It announced a plan to operationalise by Q1 2014 a public trading platform for infrastructure as service resources. The service focuses on selling by cloud providers and buying by large enterprises or intermediates such as cloud brokers.
While a cloud broker needs to understand many customer details, such as workload classification, compliance requirements and even temporary on-premise spare capacity - the Deutsche Börse Cloud Exchange (DBCE) focuses more on the trading similar to other resource trading of marketplace services, such as the electricity exchange or physical raw materials. It’s therefore more of a sourcing channel for cloud brokers than a cloud broker on its own.
Firstly, I’d like to congratulate the team at Deutsche Börse. Offering public trading of cloud resources is really a courageous and bold move. It pushes the discussion of cloud economics to the level of public trading of long term capacity, spot capacity or even futures.
Secondly – and less enthusiastic – Deutsche Börse is about to enter a steep learning curve and will most likely realise the following limitations or challenges:
Spot capacities are related to short notice sourcing. There have been other attempts to crack this market such as spotcloud.com. The new DBCE needs to partner closely with selected cloud brokers who understand the real-time demand and get the actual IT supply chain operationalised.
Long-term contracts benefit less from a real-time exchange. The sourcing of three year contracts is more similar to B2B market place trading. Cloud providers will stick to the traditional B2B – RFP process.
The DBCE demand will come from local cloud providers – not from enterprises. Many local cloud providers in Europe have built a portfolio of their own resources complemented by public cloud resources from the mega players, Amazon, Microsoft, HP, IBM, and Fujitsu. The DBCE will be an interesting sourcing option, especially for these cloud providers that will simply resell public cloud resources to complement their portfolio of self-hosted services in Europe.
Contract unification matters! Ideally customers will have only one existing frame-contract to all possible cloud providers on the DBCE. Unfortunately, German law is very difficult and outdated in terms of IT contracts. The DBCE most likely has to cover the risk of the delivering cloud provider in the sense of a supply chain. (See also the German legal situation of a IT-Erfüllungsgehilfe). If DBCE manages to keep the right balance of unification and keep itself out of the contract relationship it could work. Actually this concept is very similar to the alliances in the airline industry such as Star Alliance: unified purchasing, joint benefits such as mileage programs, but separate business units for each airline.
Reselling of spare capacity of corporate data centres is an illusion. Neither German data privacy laws, nor most cloud management stacks implemented in most private clouds would be able to share spare capacity to arbitrary external consumers brought in from the DBCE. However, Deutsche Börse could explore business opportunities around temporary spare capacities “subscribed” by European enterprises from public clouds. Some of this capacity might be up for sale and DBCE could become an independent counterpart to Amazon’s reserved instance marketplace.
Industry (de-facto-) standards matter a lot. DBCE said it would use the German cloud management vendor Zimory to offer a light level of technical services. Zimory “manage” will be used for the buy-side integration, and Zimory connect will handle the sell-side interface connecting to cloud providers. While Zimory has open APIs, it remains a proprietary cloud management solution. In contrast to this, DBCE must expect that a major portion of cloud providers work on vmWare or already on the open source based OPENSTACK and will ask DBCE to use the same “stack”.
It will only work for some workloads. IT departments or cloud brokers using a public cloud exchange basically speculate with capacity for these workloads that can be relocated easily. A forecasting calculation in the retail or financial services industry can be easily pushed to different locations every day, while a traditional ERP System will hardly change its location due to the huge data volume slowing down any movement.
- Enterprises will need to transform their business models. The engagement model between IT departments and lines of business is changing to an OpEx model. Ideally the IT department will transforms from a cost centre to a profit centre, which will enable them to cover the risk of under-utilisation or failed speculation at the exchange. Very likely we will even see a new wave of IT spin-offs stimulated by corporate cloud provider business models. However, Deutsche Börse needs to acknowledge that this is a 5 to 10 year transformation requiring long staying power.
This post is published with the permission of Forrester and first appeared on Stefan Ried's blog here.