20 Cloud Deployment Mistakes to Avoid
News | 6 May 2015
1: Believing it’s easy to migrate“The biggest mistake is thinking that cloud services are simple turnkey solutions. Just because so many of them have a ‘do it yourself’ sign-up process and may not require special hardware does not mean they are turnkey,” said Morris Tabush (@morristabush), founder and principal, Tabush Group. It’s not hard to fall into the ”it’ll be easy” trap. “Enterprises may believe that a standard application, email for example, which operates fine in their data center, will operate similarly in the cloud,” added David Eichorn (@Zensar), associate VP and cloud expert, Zensar Technologies. The reality is “there is no magic bullet that will move your assets to the cloud,” said Christopher DeMichael (@ChrisDeMichael), director, solutions architecture, RKON Technologies. “A cloud migration is a data center migration, and it brings along all of the complexities that traditionally go along with one.” Eichorn recommends testing in target infrastructure environments before any migration occurs. And DeMichael suggests a change in mindset. If you treat your cloud deployment like a data center migration, for which no one takes lightly, then you’ll be a lot more successful.
2: Not getting help to move to the cloud“Migrating to the cloud doesn’t need to be a painful process, but without forethought and third-party assistance, it can be,” said Brian Crotty (@1800Broadview), COO, Broadview Networks. “Too many organizations rely on their own inexperienced staff,” said Jeff Kaplan (@thinkstrategies), managing director, THINKstrategies. “They fail to recognize they need external help moving to the cloud.” “Cloud adoption is often a financial decision, and the cost of migration is seen as an obstacle to achieving stated savings targets. As a result, many organizations cut corners in the migration and push as much as possible to the cloud service provider who is assumed to have the requisite expertise. Providers are guilty of sometimes oversimplifying the process and ‘throwing in’ migration services that are generic, with little understanding of the applications in question,” said Sean Jennings (@virtustream), co-founder and SVP of solutions architecture, Virtustream. “This is often a recipe for massive frustration and can become a customer satisfaction issue. Experienced migration experts, knowledgeable in your applications, are worth the cost.”
3: Trying to pull off an “all or nothing” deployment“The biggest mistake when migrating to the cloud is probably to try and declare IT bankruptcy and make the move in one large, coordinated ‘swoop,’” warned Orlando Scott-Cowley (@orlando_sc), cybersecurity specialist, Mimecast. “Ideally you should think about ‘on-ramping’ to the cloud.” “Existing infrastructure and databases don’t necessarily need a forklift upgrade,” agreed Michael Krieger, (@M_Krieger), principal, MRK Technology Marketing Services. “Many companies are successful starting with an aPaaS or iPaaS [application or infrastructure platform as a service] strategy to ease a particular development or integration problem.”
4: Believing the cloud will solve your IT issues“Organizations often mistakenly surmise that a cloud deployment model will on its own solve a significant portion of their IT organizational challenges, including high CAPEX, resource waste, and slow time to market. The truth of the matter is that without a sound overarching strategy for how IT is managed, introducing cloud will only add complexity and unexpected costs,” said Jason Dover (@jaysdover), director of technical product marketing, KEMP Technologies. “Without appropriate consideration to compliance, security, performance, and an analysis of existing applications, these benefits will not become a reality as quickly as the C-suite typically hopes. To avoid this, start with an analysis of how your existing infrastructure is being used by business units across the organization, their core requirements, and the natural migration path for existing applications.”
5: Treating a cloud deployment as just a virtualization extension“The biggest mistake one can make about cloud deployment is to assume that cloud represents a virtualization extension rather than an entire[ly] different infrastructure automation model that has, as one of its software layers, virtualization in it,” said Bernard Golden (@bernardgolden), VP of strategy, ActiveState. “Organizations that make this mistake typically attempt to employ existing tools and manual processes with a cloud orchestration environment. What they commonly experience is a slightly improved virtualization environment that is inadequate to meet user expectations and demands, resulting in low utilization and an increase in direct user engagement with public cloud environments.”
6: Letting a technology-heavy perspective dominate a cloud adoption strategy“Most companies don’t go far enough with their cloud deployments. They implement the technological accoutrements of cloud-based services, but deep down they have not embraced the mindset and transformation required to attain the full business benefits of cloud,” said Jay Stephens (@orangebusiness), director of cloud services, Americas, Orange Business Services. To be seen as a valued partner to the business, Stephens recommends building cross-functional teams, determining what services you do and don’t need, and developing a plan for full system integration rather than just basic process automation. The goal is to make the entire business, not just IT, cloud ready. Once business operations can easily take advantage of the cloud, the entire organization will be ready for the inevitable and never-ending ‘adapting’ cycle. “We should now be comfortable with the reality that change will not stop, least of all the business, regulatory, and competitive aspects of building and deploying solutions,” said Thomas Fountain (@Pneuron_Corp), CTO, Pneuron.
7: Choosing a cloud service employees don’t want“In the context of internal clouds, there seems to be a false belief that if you build it, they will come,” noted Andrew Hillier (@cirba), co-founder and CTO, Cirba. “Most cloud-based solutions are built around a desire to create a digital workplace that facilitates a more open and collaborative organization. That means it’s about more than just the technology; it’s also about the users,” explained Alastair Mitchell (@alimitchell), president and co-founder, Huddle. “Businesses need to choose the cloud service that employees want to use. Otherwise, employees will end up continuing to use their favorite cloud services and will not use the company-mandated solution,” said Asaf Cidon (@asafcidon), CEO and co-founder, Sookasa. If you choose a solution without consulting users, “they will see it not as a tool to help with their work, but more as a hindrance, getting in the way of what they see as a perfectly acceptable current solution,” added Simon Bain (@sibain), CEO, SearchYourCloud. A cloud initiative is “an excellent opportunity for corporate IT to enhance its relationship with end users,” said Sam Liu (@SoonrWorkplace), VP of marketing, Soonr. “Shift from the traditional ‘top down’ application development and delivery approach to one that is more collaborative with end-users — users know what they need to be productive, IT knows what will comply with corporate policies.”
8: Overlooking the impact of your network“One of the most significant cloud deployment mistakes is neglecting infrastructure,” said Jikku Venkat (@kinvey), VP of product management, Kinvey. “Having optimal cloud compute and storage specs does not alone ensure the best possible cloud performance,” said Satish Hemachandran (@Internap), SVP and GM, cloud and hosting, Internap. “Without a reliable, high-availability network, your environment can suffer from latency, poor security, and lack of controlled access.” To build out a network that can handle transaction volume and users without compromise, Hemachandran recommends asking cloud providers questions about customer traffic segmentation, segment management, and guarantees on network speeds.
9: Ignoring the physical reality of “the cloud”“The cloud isn’t some ethereal utopia which is ‘up there in the sky,’ but that at the other end, it’s [usually very big] bricks-and-mortar data centers, which consume space, power, cooling, network capacity, and operational resources,” noted Sev Onyshkevych (@FieldViewSolu), CMO, FieldView Solutions. “Given that the underlying architecture of the public cloud is almost always powered by commodity hardware … failures should be the expected mode of operation,” warned Yuri Sagalov (@yuris), CEO and co-founder, AeroFS. “Design for failure — run your infrastructure in multiple availability zones, multiple regions, and even multiple cloud providers.” Don’t become complacent about maintaining quality and redundancy. Sagalov suggests you test for failure with Chaos Monkey, Netflix’s open source tool that randomly and unexpectedly shuts down various services and instances throughout the day and week.
10: Treating cloud resources like your physical infrastructure“The most frequent cloud deployment mistake that I see clients make is that they treat a cloud deployment like a physical data center deployment,” said Scott Maurice (@scottjmaurice), managing partner, Avail Partners. “You can easily degrade the cloud to the same level as on-premise infrastructure by applying the same old principals,” said Aater Suleman (@FutureChips), CEO and co-founder, Flux7. “Traditional metrics that measure the cost of IT in terms of the dollars spent on construction, equipment, and IT guys neglects the burden an inefficient IT department can pose on the entire business by being in the way of progress. Cloud does not just save equipment and brick-and-mortar costs; the automation abilities save more soft dollars by improving workflows and increasing productivity.”
11: Getting locked into one provider“When migrating workloads to cloud vendors, it’s easy to visualize the immediate and tangible benefits. It’s more difficult to anticipate the reasons why you would eventually leave for another cloud vendor or move workloads back on-premises,” said Jerry Sanchez (@itjerry), CIO, Planview. “Even great business relationships will end. You must go in with an exit strategy.” To prepare, Rafael Laguna (@rafbuff), CEO and co-founder, Open-Xchange, suggests you “ask yourself if your cloud needs can be met by different providers, and if there are tools to move from one provider to another.” “When your new production app built inside one cloud actually takes off, you will want the option to move it or expand it to other cloud services, and even be able to bring it back on-premise,” said Mike Matchett (@smworldbigdata), senior analyst, the Taneja Group.
12: Avoiding hybrid solutions“Don’t think about cloud deployment alone,” suggested David Poarch (@DavidPoarch), VP of security solutions, Forsythe. “Focus on the best way to operate a hybrid environment, seamlessly incorporating both legacy and digital cloud capabilities.” “Deploying to hybrid cloud services,” said Ian Apperley (@ianapperley), writer and IT consultant at whatisitwellington, “allows greater flexibility, greater cost control, increased functionality where you need it, when you need it, and much lower risk.”
13: Not thoroughly vetting cloud solutions“Price and performance between clouds vary greatly,” warned David Cope (@DavidJamesCope), EVP of corporate development and CMO, CliQr. His company’s app portability and benchmarking tool has seen 50x differences between cloud environments. While it’s hard to do an analysis of cloud providers “that’s no reason to just give up and go with your gut, or with whomever is most popular, or with what your friend Joe who’s pretty smart says you should do,” advised Scott Feuless (@ISG_News), principal consultant, ISG. “Choosing any cloud solution without respect to your particular business’s needs would be a mistake,” said Mark Shirman (@RiverMeadow1), president and CEO, RiverMeadow. There’s a long list of features or lack thereof that can bite you, said Feuless, if you don’t evaluate them with respect to your business’s needs beforehand. Those include, but are not limited to, portability, elasticity, support, compliance, auditability, SLAs, governance processes, scalability, reliability, fault tolerance, recoverability, invoice predictability, NPV, interoperability, response times, security, and privacy. Once you narrow down your list, said Rusty Putzler (@rputzler), VP and COO, Connectria, “take a visit to your provider’s facility to actually see how it operates. If they are reluctant to schedule a visit, consider it a huge red flag and move on.”
14: Poor cloud deployment planning“People make mistakes moving to the cloud when they disregard their current environment. Moving unstable or stale data from your data center to the cloud creates stale and unstable data in the cloud,” said Tim McKellips (@McKellip), manager of technical services, Softchoice. “It’s easy to lose track of where you are and only focus on where you’re going.” Many experts we spoke to warned of the detriment of poor cloud deployment planning. For example, if you don’t take the time to fully understand your unique business and industry compliance risks, the result is “many deployments become vulnerable and the owners experience more risk than necessary,” explained Jeff M. Spivey (@spiveyjms), VP of strategy, RiskIQ. “The deployment ends up costing more due to the reactive application of security as opposed to planning full integration from the beginning.” Sumeet Sabharwal (@sabhas), group VP and GM, NaviSite, adds that “the journey to cloud begins with a sound strategy up front, one that maps out the key drivers, the associated business case, expected outcomes broken down by milestones, and a phased migration plan.”
15: Choosing the wrong class of service and too much of it when deciding on whether to choose an IaaS-, PaaS-, or SaaS-based cloud solution, “too often we see a ‘build first’ mentality instead of ‘solve first’ mentality,” said Leo Reiter (@virtualleo), CTO, Nimbix. “For example, instead of choosing a provider who can deliver the application to get the job done, consumers often resort to just spinning up ‘instances’ and trying to install, configure, and manage the software themselves. But if you're just trying to solve a specific problem, you should really be looking at turnkey solutions delivered in the SaaS model.” When decisions about cloud solutions are made ad hoc, you’ll run into a situation of trying to manage multiple distinct cloud instances. “Organizations need to resist the impulse — understandable though it may be, given the ease and affordability of individual solutions — to get into lots of disparate clouds, with disconnected applications,” said Adam Stern (@iv_cloudhosting), CEO, Infinitely Virtual. “‘Do I really want my data all over the place, without controls?’ That’s the very definition of cloud sprawl.”
16: Underestimating costs“Failing to do proper cost planning often results in unpleasant surprises when the bill shows up,” said Gerardo Dada (@GerardoDada), VP of product marketing and strategy, SolarWinds. “The lure is understandable. It’s easy and inexpensive to put a lot of data in a public cloud. So much so that many organizations adopt cloud solutions at the department level, adding to [them] progressively over time,” said Jeff Flowers (@Storiant), CEO and co-founder, Storiant. ”If there are no proper processes in place to monitor and control cost, it can run away from you very quickly,” warned Koos du Preez (@K2onK2), CTO, K2. “Companies can avoid the mistake [of cost overrun and data lock-in] by centrally planning and implementing a private cloud strategy that delivers the same scalability and economics of public cloud, but without the liabilities,” Flowers continued. In addition, Max Dufour (@maxdufour), partner, Harmeda, recommends you “evaluate solutions beyond just cost considerations and based on actual business value. It makes a difference to thoroughly analyze benefits and features rather than target marginal savings on initial costs.”
17: Setting it and forgetting it“When it comes to cloud services companies often think of the cloud as an arm's length relationship with an unknowable company. This lack of communication is a big mistake, especially if you place mission-critical services in the hands of the cloud provider,” said Dan Carney (@llnw), VP of operations, Limelight Networks. “Too many companies rely on the cloud, but don’t monitor and measure the performance of their cloud services on an ongoing basis,” said Matt Larson (@Matthewhlarson), CTO, Dyn. Adrian Sanabria (@sawaba), senior analyst, enterprise security practice, 451 Research, recommends you “don’t use a cloud vendor unless they give you some way to effectively monitor console access and activities.” Degradation of cloud performance does happen, and you won’t be able to rectify if you’re not monitoring and mitigating issues as they occur, said Larson.
18: Ignoring disaster recovery planning“Traditional disaster recovery planning is still just as important as before, because cloud-based data centers can still be impacted by disasters,” said Denny Cherry (@mrdenny), owner and principal consultant, Denny Cherry & Associates Consulting. It may be important, but the overwhelming majority of cloud users are not preparing for the worse even though their on-premise network is just as susceptible to disasters as the cloud. According to respondents to a recent Vision Solutions survey, nearly two-thirds of organizations admitted to not having high-availability or disaster recovery protection in place for their data once it’s in the cloud. For true DR recovery, you need off-site backups, not just another instance with the same cloud provider in a different location. “If either the enterprise or the cloud vendor suffers an identity breach, an intruder could delete all of the enterprise’s workloads and backups in every cloud region, typically through a single console and login ID,” said Charles Moore (@delphix), product marketing, Delphix, who notes this scenario happened to Code Spaces, which was forced to shut down.
19: Not automating everything“Biggest mistake is not automating everything,” said Dave Nielsen (@davenielsen), co-founder, CloudCamp. “Automate pretty much everything, such as spinning up a VM or container and then updating the configuration files as required by the application. If a sysadmin finds themselves doing the same task over and over again, it should be automated.”
20: Only looking at cloud as a short-term strategyAsking “What can the cloud do for me right now?” is a short-sighted view on the revolutionary impact the cloud can have on your business. “Cloud is really more of a change in business strategy than in IT functionality,” said Meir Uziel (@clarizen), VP of managed services, Clarizen. “The cloud allows businesses to consume IT resources as a service, which has a lot more reach into process than into actual technology and requires a deeper connection between IT and a company’s business units as a result.” Conclusion: Don’t jump in without setting yourself up for successThat’s a very long list of mistakes (we actually received more), and it may give you pause to even consider moving to the cloud. Don’t fall into that trap. “The single biggest deployment mistake most organizations make is waiting,” said Tim Cuny (@OptimizewithCMI), VP of solutions, CMI. “To set yourself up for success, define the goal and scope of the initiative,” said Lauren Nelson (@lauren_e_nelson), senior analyst, Forrester Research. “Dive into the economics of what you're paying for and how you and your peers are incentivized to use the product.” If you have experience deploying the cloud and have even more suggestions on what people should avoid, please add your thoughts to the comments. If we all learn from each other’s mistakes we’ll improve the overall success of cloud deployments.