Issue link: https://cp.revolio.com/i/584904
24 | THE DATA CENTER JOURNAL www.datacenterjournal.com t he cloud offers many advantag- es to businesses. Cloud-based infrastructure-as-a-service (IaaS) provides flexible, on- demand, consumption-based IT options for organizations of all sizes. Cloud environments are characterized by lower operational costs, drastically reduced capital expenditure and greater availability, and they allow IT staff to focus on their company's expertise, as opposed to the mundane tasks of infrastructure upkeep. cLouD ProviDers coMPareD Two types of IaaS cloud-service pro- viders are in the marketplace: hyperscalers and enterprise-class private-cloud provid- ers. Let's define the two provider types by their consumption model. Hyperscalers sell cloud in "by-the-instance" increments. e instance is a preset and pre-sized, (i.e., S, M, L or XL) virtual chunk of memory, processing cycles, storage and so on. Hyperscale-provider deployments are handled by companies with developer- centric IT staffs that require little if any operational and management assistance from their provider. For companies that need only to scale horizontally, as well as companies with read-only applications, hyperscale cloud providers and by-the- instance consumption models offer cost- effective and straightforward solutions. In contrast, enterprise-class private- cloud providers sell cloud in "resource pool" increments—no t-shirt sizes here. Instead, every element in the cloud's resource pool is fine-tunable on demand, including RAM, CPU, storage, IOPS and network bandwidth. is approach is suitable for complex environments such as transactional systems that require high availability (HA) and custom fine tuning of compute, network or storage indepen- dently of one another. Availability design is built into the lowest layer of the stack, and it can support both horizontal and vertical scaling. It's typically not the easiest IaaS model to deploy, but it is the most customizable provider type thanks in part to the granular level of fine tuning avail- able on its resource pools. Hyperscale clouds tend to be less expensive on the front end. On the back end, however, they can prove to be far more costly should customers require customization or add-ons such as HA redundancy, as well as seeding, retrieval and frequent movement of data. Enter- prise private clouds are oen costly at first, but in the long term they provide savings if a company desires customization and fine-tunability and has ever-changing data stores. consiDeraTions When Moving To a ThirD-ParTy cLouD-service ProviDer Knowing that they have two distinct alternative cloud-provider models to choose from means companies must be vigilant in their due diligence. Every pos- sible area of business impact should be considered before executing a migration. Considerations include the following: 1. Service disruption: To what degree can you afford service disruptions during the migration process? Do you need high availability, or can you afford to run my business in a non-HA environment, with possible interruptions? 2. Company agility: Are company resources available during the process? Does your cloud provider employ the same virtualization tool set that your resources are already trained on? If not, what is the cost to retrain? 3. Political fallout: Can you meet your timelines using your existing corporate resources? What are the political implications if the team cannot meet the deadline? What if the cloud envi- ronment encounters downtime? 4. Financial impact: One cloud model is cheaper on the front end; the other may be cheaper in the long term. Given the complexities of your environment, it's horizontal or vertical nature and transactional versus read-only characteristics, which cloud type makes better financial sense in the long term? Once you recognize these main risk areas, you can develop a strategy that bal- ances your risks with the executive plan. is strategy can be broken down into four steps: Step 1 - Define and track key performance indicators (KPIs) for the cloud migration Step 2 - Gauge the organization's appetite for the business impacts and risks above. Are the risks negligible? Is the organization's migration plan technically viable? Hyperscale clouds tend to be less expensive on the front end. On the back end, however, they can prove to be far more costly should customers require customization or add-ons such as HA redundancy, as well as seeding, retrieval and frequent movement of data. Enterprise private clouds are often costly at first, but in the long term they provide savings if a company desires customization and fine-tunability and has ever- changing data stores.