Article

Workloads and the cloud: To host or not to host?

The cloud has taken enterprise IT by storm in such a short period of time that a lot of organizations either have no idea they're leveraging cloud services or they use so many SaaS and IaaS solutions that they can't even count. It's common for business leaders to simply hear about the benefits of cloud computing, and then it's off to the races for those companies as they try to deploy workloads on hosted infrastructure.

Unfortunately, as more organizations rush to implement and take advantage of cloud-based applications, hosted hardware or infinite Internet storage, they make small mistakes and can potentially knock their journeys off of the rails, causing them to lose time and money as they clean up the mess that's left. If a migration to the cloud fails, not only is there a chance to lose mission-critical applications, but business leaders will most likely call off the whole process as well. This will ultimately hurt the organization, as the cloud is required to remain competitive in the modern era. The bottom line is that a company cannot expect to grow without implementing cutting-edge solutions to problems that have plagued them for years, including but not limited to flexibility, scalability and resilience.

Therefore, every attempt to capitalize on the cloud should be well-planned and thorough in scope. Simply put, not all workloads will live happily in cloud-based environment, yet at the same time, there is always an approach for organizations to follow when deploying business-critical applications and data in hosted data centers.

'Two-speed IT'
One strategy for the cloud would be to first separate internal IT demands - in regard to employees using tech - and external requirements - those facing customers and clients. TechTarget reported that John Deere takes this approach to enterprise IT, and they call it "two-speed IT."

Richard Johns, a solutions architect at Computer Sciences Corp., explained that the traditional infrastructure that keeps a company alive should receive just as much attention as bringing products to market, but they require different underpinning strategies.

"Internal IT projects deserve longer cloud migration cycles."

For example, getting consumer-facing cloud-based services off the ground with speed should be a high priority, while internal IT projects deserve longer cloud migration cycles because if those don't work properly, business will suffer as a result.

At the end of the day, Mano Mannoochahr, director of enterprise architecture, information management and computer security at John Deere, reminded TechTarget that decision-makers should ask themselves whether the workload is helping the core business or creating a new opportunity to provide for their customers.

Ease of deployment and potential gain
In essence, the cloud isn't for every workload due to the fact that some applications, databases or tools don't need to operate at different scales depending on the time. When looking at what belongs in the cloud, IT admins should consider the intangible benefits - such as customer satisfaction - alongside the actual value that hosted environments bring, according to IBM. Some workloads will deliver more gains than others, and the source recommending weighing those pros with the cons of deployment. Specifically, workloads with little to no customization required will easily migrate to the cloud.

IBM created a chart detailing which workloads will be easiest to deploy compared to the potential gains they will provide. Development/test environments topped the list in regard to the number of benefits and the highest degree of simplicity. On the opposite end of the spectrum lies complex ERP systems. Closely behind development environments, IBM listed collaboration tools such as hosted PBX and unified communications and storage solutions, which could include AWS S3 or a similar service.

Private vs public vs hybrid
The type of cloud environment will also determine which workloads will function the best on hosted infrastructure compared to local data centers. Open Group made the argument for putting workloads in private clouds stating that those are attractive when security and privacy are major concerns. However, mitigating those problems introduces scalability limitations, so big data analytics in particular wouldn't make sense in that type of environment.

Workloads belong in the cloud when they can easily migrate and provide plenty of benefits.Workloads belong in the cloud when they can easily migrate and provide plenty of benefits.

Open Group suggested that the public cloud is great for workloads that are temporary. This allows for elasticity and the appearance of infinite capacity that is necessary when testing applications.

Then there's the hybrid approach. This is probably the best, as data can reside on premises, yet applications and development environments can take advantage of unlimited computing resources. As another example, Open Source noted that business intelligence solutions can live on hosted infrastructure and data can slowly migrated and removed as necessary.

Other considerations
There are plenty of considerations when looking at which workloads belong in the cloud, and there really isn't a one-size-fits-all approach. Decision-makers should forget to consider APIs, costs, compliance and bandwidth requirements. Sometimes an extra solution such as Direct Connect or bandwidth on demand are necessary components of cloud computing environments, so IT admins should be sure to create lists of their demands and needs.

John Grady is a Senior Product Marketing Manager at XO Communications, which is responsible for marketing the XO Cloud and XO Connect product portfolios. John has been responsible for launching a number of products at XO Communications, including 100G service and new cloud products, as well as XO's Intelligent WAN solution.