Despite the hype many companies are resisting a major cloud transition and this is often at the recommendation of internal IT managers. So the question must be raised – why? While every company is different I see four primary sources of resistance and each source has a separate rationale. From CIOs to IT managers, system engineers and system architects, there are both valid and invalid concerns:
|CIOs||Cost – Operating versus Capital|
|IT Managers||Resource Retention|
|Architects||Data Security, Portability, Reliability, Data Loss, Performance|
As reported by CIO Journal, CIOs are split 50/50 – due to a cost perspective. Simply put, an up-front hardware expense is a capital cost whereas an ongoing cloud subscription is an operating expense. Mature companies view capital costs as ongoing innovation whereas operating costs take away from the bottom line and that looks bad. Capital costs can be throttled back in preparation for a divestiture or as a temporary cutback in times of economic challenge but operating costs are a hard dependency. By migrating to the cloud you are making a long-term commitment.
This is a valid concern but one has to weight the total annual operating cost against: (1) the revenue generated as a direct result of these systems, and (2) against the depreciation of local hardware assets and the cost of the professional human resources required to support them. In my own experience I can run 30 production servers in Amazon Cloud for less than 35% the cost of a single Systems Administrator. As a result my operating costs are actually still cheaper and my capital costs are restricted to “server reservations” (up front payments to reduce the monthly costs of those resources in exchange for a longer-term cloud commitment), which still equates to far less than the cost of traditional hardware.
IT Directors and Managers
Any good IT director or manager is not only concerned about business growth but also about the well-being and career development of his direct reports. But, at the end of the day the business needs trump individual needs. IT managers know this and I think have a tendency to push back on the change in fear of having a negative impact on the IT organization; that is, IT managers fear that migration to the cloud might result in job loss. That to me is indicative of ineffective management – lacking either technical knowledge or a sound management strategy (as I mentioned in a previous post, “4 Ways CIOs Can Immediately Leverage the Cloud”).
In actuality the benefits of the cloud actually better the image of the IT organization as they create more efficient use of resources at significantly reduced cost and with far greater flexibility. I could elaborate on this topic in 100 different ways but I’ll await your feedback in order to better focus that topic. Essentially, IT managers need to approach the cloud as a tool for increasing efficiency and not interpret it as a means for reducing resources. Servers still require maintenance and patching and applications still require an architect – preferably one that can leverage the suite of tools offered by various cloud services.
The IT industry is volatile and that is no surprise to anyone in this industry (if it is you should seek a fall-back option immediately). Technology changes quickly – software, hardware, programming languages, and “best practices”. If you can’t keep up with the changes then you will fall behind and quickly find yourself restarting your career as a level 1 engineer because your specific knowledge is no longer relevant. IT engineers I think develop a natural resistance to change because they tend to view change as an indication that they are being phased out. Leadership needs to provide a clear road map in order to alleviate these concerns and offer training to facilitate this transition with existing resources. The engineers are more likely to support it and in fact they’ll probably relish the opportunity to expand their own skill-set.
On the other hand hardware engineers are concerned about the loss of physical access to the data center. Engineers are not comfortable if they cannot touch things and if the data center is not across the hall or in an adjacent building then they have no control when the physical hardware fails. This is a valid concern and indeed, when your cloud hardware fails (and it will), you will not have access nor can you call someone who does. This is a concern that has to be addressed in your architecture itself – and must be well tested prior to Production deployment. Data storage must be redundant and your servers (hosting your applications independently from your data stores) are simply an interface for working with that data. If a server fails you simply launch another one and if you architect your systems correctly this happens with zero impact to the applications themselves.
I still cannot decide whether architectural concerns are actually coming from the architects or being raised by IT managers attempting to defer a cloud transition. Either way the rationales are certainly architecturally based and so I am categorizing them as such and labeling the architects themselves as the source. Architects tend to have the following concerns – all valid (not in any particular order):
1. Data Security – Are my customer’s credit card numbers safe in the cloud? You can certainly encrypt your connections, data volumes, even configure a Virtual Private Connection (VPC, Amazon’s answer to a VPN) in order to protect your data stores and connections to/from that data. In reality your data is no less secure than the transactions you’re already processing online and I would suggest it is more secure than hosting internally – where “trusted” employees likely have some level of access to that data and security requirements are likely more relaxed because the internal network is viewed as a “safe place”.
2. Portability – Can I move my applications and data elsewhere if need be without significant challenges? This depends entirely how you choose to utilize the cloud and architect your applications. For example, if you are using the cloud for off-site data backup then this is probably not a rational concern. If you choose to build your application around a specific proprietary cloud service (e.g. Amazon CloudSearch) then, yes, you will need to account for that and be able to replace that functionality if you opt to transition away from that service in the future. Isolate each component of your application and use domain-driven design (DDD). If you don’t want to limit yourself to a specific cloud offering (from a design perspective) then consider a service provider such as RightScale.
3. Reliability – IT service providers today have gained trust by providing rock solid SLAs and today’s cloud providers have not taken that same approach. Reliability is a concern and we’ll have to wait and see how these SLAs mature. I view the existing SLAs as a foundation that will encourage the growth of middle-management companies that will resell major cloud services with a value-add SLA on top of a guaranteed management/maintenance agreement. Until then the reliability factor has to be considered per use-case. For your corporate website the reliability might not be the primary concern but I might not put my VoIP system on the cloud just yet if it will mean that business communication comes to a halt when there is a hiccup in your cloud service.
4. Loss of data – Remember, “the cloud” is not this imaginary world of computers. It is a collection of physical data centers strategically placed geographically. There are numerous levels of redundancy (e.g. Amazon EBS, S3, Glacier, and reduced redundancy options). While data loss continues to be a primary concern, note that the causes of most data loss are no different (although they might be more frequent) than in a local data center. So long as you draft and implement a data recovery plan and suitable backup procedure this is not a valid reason to avoid the cloud. My current environments leverage daily snapshots of all data volumes (100% coverage), machine images of all running Production instances (backed up to Amazon S3), and one weeks worth of daily full database exports (contained within the EBS snapshots) and, all automated. The next step is to duplicate these in an entirely different region (just as companies eventually grow into a global load-balancing solution I do not view this as an initial requirement).
5. Performance – Performance can be measured in many ways so as always individual needs have to be considered. In general you can leverage the cloud as a CDN (see Amazon CloudFront), configure EBS volumes for high IOPS, etc. I do not have any experience with data warehousing so I cannot comment on that but with regard to web hosting performance is typically above and beyond any small to mid-size companies’ current capabilities. Even if you observe decreased performance in local I/O you’ll likely make up for it in content distribution. And if not, you can easily scale up your resources at low cost to account for the degraded performance. Again, personally I’ve observed that a properly architected environment has no noticeable performance degradation and I’ve been able to leverage content distribution solutions that were previously out of my price range with companies like Akamai.
Resistance is a natural reaction to a new technology but there are ways to alleviate these concerns in most situations regarding a transition to the cloud:
- CIOs need to view the total cost of a local data center in comparison to the cloud – management, maintenance, electricity, etc. and take into account hardware depreciation.
- IT managers need to communicate a business road-map and provide training opportunities to engineers in order to alleviate concerns and encourage a smooth transition.
- Most of all architects need to realistically evaluate their individual needs, understand the new SLA format, and design systems that account for the new risk factors.
- How are people using Cloud Computing? (cloudave.com)
- Financial services and the public cloud: Go or no go? [GigaOM] (gigaom.com)
- New Independent Research Study on Hybrid Cloud Business Trends and Technical Challenges (blogs.cisco.com)
- Cloud Adoption Is All About Value for CIOs (sys-con.com)
- Netpremacy announce 5 simple steps to secure corporate data in the cloud (prweb.com)