Previously, I had dealt with some of the complexities of computing as it exists within enterprises today, as well as intrinsic aspects of these workloads. These are important factors that have a great influence upon the decision to migrate them to the Cloud. Earlier posts can be viewed here and here. In addition, there are a number of factors that inhibit rapid Cloud adoption by customers. Some of these are
1. Standards, Standardization, and Interoperability between Cloud providers
Today, we have multiple providers selling *aaS services within their public Clouds, and with products for Private Clouds. Each vendor is attempting to stake out new ground with regard to features and capabilities. There are some basic similarities between these clouds in terms of support for features, but they do not interoperate with each other. Migrating a workload from one Cloud to another is a lengthy and manual process, with multiple steps involved, making it only useful as a one-time migration activity. This process does not permit automation, and significantly increases costs and risks for organizations.
2. Vendor lock-in
While there are multiple providers offering similar services, each of these vendors is interested in acquiring new customers, and then locking them within their Cloud forever. Amazon makes it easy and inexpensive to migrate in-house workloads to their Cloud, but makes it very difficult and expensive for customers to leave. Other vendors utilize similar strategies through a combination of product features and pricing. Customers have experienced the impact of vendor lock-in on many occasions, with several products and vendors; hence they are wary of this vendor strategy.
3. The rip-and-replace issue
A large organization has many applications that generate data, communicate, and interoperate with each other via automated workflows. En-masse replacement of all applications is not feasible, and needs to be accomplished one application at a time. Each application migrated to the Cloud needs to be reconfigured to communicate with the desired in-house applications. There is significant expense and risk involved with nearly zero or very low ROI until the entire migration is complete. Even at that point, the potential ROI is unclear.
4. The scale-up paradox that accompanies growth
We dealt above with the Cloud adoption curve – as a new startup grows, its variable costs increase. The complexity of its applications and the number of providers also increase. At some point, it becomes feasible and necessary to consider investing capital to reduce the variable costs, for at least the most expensive of its providers. This would be contrary to the direction of the industry. In addition, the ROI is quite clear for organizations with large, low-scope applications, and for small startups. When such a startup grows to become a mid-sized company, both scale and scope increase, with the recommendations being unclear over a wide range in the middle. Also, is there a strategic reason to bring it in-house sooner?
5. Buy vs. rent
The main selling point for a public Cloud service is that large capital costs are instead converted to manageable operational costs. This is attractive and meaningful for small companies, where it is easy to acquire a service without a need to acquire a new skillset in-house. However, Cloud services are rented by the hour, and by the capacity tier. Given a choice, would you buy a car, or rent it on a daily basis when the plan is to use it every day over the next 5 years? What if you needed it only for the next week? The recommendations for these cases appear obvious; what about moderate usage?
Cloud services represent capital costs for the provider, who must recover them by renting this capacity at higher prices to customers. With a relatively stable workload, it might be feasible for organizations to establish in-house systems, using the public Cloud for overflow capacity in a Hybrid model.
6. Risks of working with startups that could vanish
There are a number of startups that offer Cloud-based services that have the potential to replace in-house legacy applications. Examples include ERP, Databases, (list some more). It is quite likely that they do not offer all the features, or the sophistication of legacy software. Enterprises might consider migrating such applications with the understanding that more features could be added down the road. However, there is an inherent risk of the provider going out of business, which can be very sudden, causing serious impact to its operations. This is one of the reasons that established companies adopt a cautious approach to migrating critical applications.
7. Performance and network latency considerations
Cloud based providers operate multiple datacenters, which can be local (low-latency) to the enterprise, distributed, for a multi-location organization, or remote to aid Disaster Recovery. Each of these has advantages and disadvantages, and is based on the end customer requirement. Since a single provider is unlikely to be able to fulfill the needs of many mid-size and large organizations, it will be necessary to utilize multiple providers in a Cloud-based operational scenario. A multi-provider scenario would also be recommended to avoid down time resulting from a Cloud provider failure.
Now, there are provider to provider latencies to consider as well; if each provider were to use different telecom providers, then the landscape gets pretty complicated to implement and manage. Given the difficulty enforcing SLAs and OLAs with any single provider, a failure of communication between providers is likely to lead to finger pointing between them; in the meantime, the organization and its customers are faced with sub-par performance.
8. Long-term normalization of Cloud Costs
Companies such as Amazon, Google, Microsoft, and many others have been racing to establish datacenters and tout these capabilities to attract new customers. They have been continually lowering their internal costs as well, attempting to match each other’s price reductions. This is a strategic move by large companies generating significant profits from their dominant primary business to squeeze out new entrants within this space (Rackspace?), and also to establish their design as the prevailing standard. While costs for customers are low today, it is primarily due to this factor. We do not think that either of the large providers are realizing commensurate profits based upon their investments towards these infrastructure developments.
Also, many of these infrastructure projects have been generously subsidized by governments, and take advantage of low costs for land, electricity, and manpower in the locations of these datacenters. These incentives will run out and other costs will rise over time. In addition, these providers could come under pressure (either due to a focus upon generating ROI for their Cloud investments, or due to reduced profits within their primary business), and will need to pass on some of the costs to customers. Such price increases could create significant uncertainty about the expected ROI for companies that are planning to migrate to the Cloud.
9. Outdated Business Models
The world of economy and business has evolved drastically over the past decades. Some examples of this include globalization, use of the Internet, functional (instead of hierarchical) organizations within a flatter structure, greater integration with partners, leverage from social media, and much more. However, the IT systems reflect the vestiges of older business models, and constrain the business processes from change. While Cloud-based services reflect today’s business models, they represent drastic change for many organizations. Established organization within the manufacturing, financial, or government sectors are highly prone to this issue, while it is of much lesser consequence for newer companies in other sectors.
In this context, updating business models is imperative in order to effectively leverage cloud services. It is to be noted here that piecemeal adoption will provide lesser benefits until the business models are updated. This is a significant opportunity for established organizations and companies that successfully transform could represent a serious threat to the holdouts. In any case, newer and agile companies within this space will reap the initial benefits from Cloud services merely from the absence of legacy systems.
Cloud Computing consists of a relatively simple set of tools and technologies, and are expected to make rapid progress towards the Plateau of Productivity within Gartner’s Hype Cycle shown below.
Hype Cycle for Emerging Technologies, 2014 (Gartner, August 2014)
However, both the timing and ultimate scope of Cloud Computing are subject to a number of factors. A failure to achieve these could extend these time frames and/or reduce the viability and adoption of Cloud Computing. These factors include
Vendors need to offer modular services that permit customers to select features that they need, without having to pay for those that they do not. In addition, these should work with legacy protocols that customers already use for in-house systems. This will permit incremental replacement of legacy in-house modules with Cloud-based replacements and will not need extensive customization or integration. Once customers are comfortable with the concept and ease of module replacement, this process would accelerate.
Any effort to address the disjointed landscape of Cloud-based services will increase the comfort level of customers. At present, customers hesitate to even consider deploying systems that utilize multiple Cloud vendors’ services, solely due to the hurdles involved with integration, as well as the time and effort involved. The technology workforce and consulting companies have expertise with multiple vendor services today; yet they lack the skills to effectively integrate these services seamlessly. This is a significant barrier to companies that wish to better utilize compelling services from competitors within their application environments. Companies may also wish to arbitrage their costs of Cloud-based services by utilizing multiple vendors, with changes being static or dynamic. Providers offer dynamic pricing based upon their current resource utilization, and it would make sense for customers to leverage this.
- Reduction of Cloud and Network Costs
While the cost of Cloud services are on the decline today, much of this is due to the large vendors staking out positions to establish themselves as the market leader, and to deter new entrants into this space by temporarily reducing market profitability. Long term costs of these services should therefore rise to match economic costs; some of this would be offset by progressive reduction in technology costs. Unless customers see a permanent reduction in costs, with a falling trend steeper than that of their current technology costs, it would be difficult to get medium to large-sized customers to adopt for mainly cost-based reasons.
Network costs are an important part of this mix as outlined above; in case network charges drop steeply due to technology improvements, or from an increase in competition within this space, customers that move large amounts of information would have a greater appetite to adopt Cloud-based services. This is likely to be attractive to them, since they are cost-neutral at worst, and provide additional benefits.
Customer-centric Cloud offerings
Today, we have vendors that offer services that represent hardware systems, technology platforms, databases, or other services, all of which represent technology services. On the other hand, customers utilize and need financial, logistic, manufacturing, engineering, and other similar services, which can be categorized as operational services. It is almost as if customers and vendors exist in their own universe. This is probably one of the most critical factors that currently restrains adoption. Unless the industry offers services that are focused to customer needs, adoption is likely to be muted at best. Notable exceptions here are the offerings from companies such as SalesForce and ServiceNow.