One of the big items in technology news this month was the Netflix announcement that it would be powering down the last of its datacenters, bringing to conclusion a multi-year Cloud migration initiative. Adrian Cockroft, the head of the infrastructure team at Netflix who managed this migration, predicted that more enterprises would adopt a Cloud-only initiative. There are many experts that claim that the corporate datacenter is dead, but is that the reality?
At the same time, SalesForce has been busy establishing new datacenters all over the world. This is also true for other players such as Microsoft and Google. The stated advantages of the Cloud are lower costs, simpler setup and operation, and easier management; establishing datacenters does not seem to make any sense. Is the Netflix model universally applicable to all companies, or is there more here than meets the eye? Let us first take a brief look at what worked well for Netflix.
- Business Model: Netflix offers movies and TV shows on demand to subscribers, which is downloaded for viewing on PCs, televisions, and other mobile devices. Subscribers are likely to be indifferent to the source of their content since their actual interface and experience will not change. Not many companies have such a limited range of operations and applications.
- Impact of network costs: This factor is highly favorable to Netflix as the content moves closer to the consumer. It reduces both operating costs and download latencies. Most large companies have legacy applications, and will find that their network costs would drastically increase with some applications residing in the Cloud. They could also see increased latency with legacy apps being run in the cloud.
- IP considerations: Intellectual Property (IP) considerations are paramount when companies consider moving their crown jewels to the Cloud. However, the content in this case does not belong to Netflix. The software application could be considered their IP; however even this could be treated as replicable. The major IP for Netflix arises from their people and processes, such as their rating and recommendation systems. This information still stays in-house, so there is no risk from Cloud migration. However, companies such as Intel would prefer to keep their cutting-edge chip designs in-house.
- Externally generated data vs. internal: Most of the data generated at Netflix arises due to subscriber interaction with the application. It is externally generated, in contrast to the internal data generation of manufacturing organizations (such as GE) or research focused organizations (such as the Pharma industry).
- Regulatory and Compliance limitations: There are no regulatory or compliance limitations upon Netflix’s business model. Their liability arises from the privacy of their subscriber information and is the same in the Cloud, as it is in-house. Companies subject to regulations such as HIPAA or ITAR are restricted to compliant providers. They will find that choices are limited and would be quite expensive.
In addition, the Cloud provides Netflix with increased scalability and elasticity. Since they receive monthly subscription revenue, converting their capital expenses into operational expenses provides them with better financial visibility. This was likely a significant factor in Netflix decision.
Before we rush out to compare and shortlist Cloud providers, it is prudent to evaluate our own organization and operations. A simple method is to answer the questions below with a Yes or No
- Are your workloads relatively consistent (irrespective of season, product/market cycle, etc)?
- Do your workloads generate a high volume of data?
- Is most of your data internally generated?
- Do your workloads need to interact with legacy data or legacy applications?
- Are there IP protection considerations that prevent migration of sensitive data?
- Is your company or industry subject to regulatory or compliance limitations?
- Would your network costs increase significantly with migration to the Cloud?
If you answered “No” to all these questions, many congratulations. You are ready to move your applications to the Cloud, generate a shortlist vendors and get quotes to identify your ROI.
In case you answered “No” to most questions, there may be certain workloads that could move to the Cloud while others need to remain in your datacenter.
If you answered “Yes” to most or all questions, it is time to reinvest in your datacenters for some more time.
A note about size of the company and scale of application workloads: while these need not always be correlated, smaller companies have smaller–sized workloads, which grow with the company. The economics highly favor the Cloud for smaller companies. However, this advantage diminishes with growth of the workload, as operational expenses scale up rapidly.
It makes sense for very large workloads to be retained in-house; all the large players including Google, Microsoft, Facebook, and Amazon are focused on building, maintaining, and expanding their datacenter footprint. This means that at some time during the growth phase of a company, Cloud-based workloads might need to be migrated to in-house datacenters for purely economic reasons. Strategic reasons might further speed up this process. For most companies, a hybrid operation would make the most sense; migrate some applications to the Cloud for cost and flexibility reasons, while retaining others in-house for other reasons.
Still confused? The reasons to migrate to the Cloud and the reasons to keep workloads in-house will continue to evolve, making this a difficult and subjective decision for many companies. There are also many barriers that inhibit companies from quick adoption of Cloud computing. I dealt with this issue in detail last year; while there have been some improvements, most of the issues continued to be valid and unaddressed by the major providers. Please let me know your thoughts on how you see this evolution progressing and what the industry needs to do to speed it up.