Among the various trends that I have observed over the last 12 months of working smack in the middle of cloud planning discussions, I would like to address one that seems counter-intuitive at first blush but is a definite shift in how forward-thinking enterprises are planning for the cloud.
Enterprises need to reverse their order of priorities when planning for cloud.
There are 3 priorities that enterprises weigh during cloud readiness and migration assessments of their on-premise applications – Security, Performance and Cost – usually in that sequence.
Most enterprises typically evaluate security and performance concerns in detail, short list a public cloud platform, and then estimate the costs and perform the migration. However, in order to identify the most optimal placement for their applications, enterprises need to reverse their order of priorities – estimate the costs, then evaluate security and performance considerations. This might appear to be a controversial statement but consider how much the cloud landscape has changed in the last 12 months – cloud platforms have gotten more mature and a multitude of provisioning/migration tools have come into play. Companies such as Racemi, Sureline Systems, CloudEndure, ATA Data and CloudVelox offer “lift & shift” migration of applications and data while others such as CliQr, Gravitant and Avni ease provisioning within a hybrid management tool set. Enterprises have a choice of tried and tested tool sets that can move on-premise applications to the cloud.
Security and Performance Considerations are roughly equivalent across the Mega Clouds. Cost is the Differentiator.
The battle among the mega clouds has never been more intense. Read this USA Today article that summarizes the momentum and market share of the current behemoths and challengers. All the mega clouds – AWS, Azure, GCP and Azure have added data centers across the globe – not just in North America but in Australia, Asia and Latin America. Local access to applications reduces latency and improves application responsiveness. Nearly all cloud vendors offer solutions for big data applications with blazing analysis speeds. Beefed up security features, both native to the cloud vendors, and offered by a plethora of partners satisfy the need of the most security-sensitive organizations. So, security and performance options are roughly equivalent across all the mega clouds. Costs, however, vary widely across the clouds with great variance in charges for services such as compute, storage, memory, network bandwidth, IO transactions and application licenses. So it makes sense to evaluate the differentiating factor – cost – first before pointing the lens at security and performance.
Cost Modeling Remains Complex.
Calculating the costs of running an application in the cloud is not a trivial task. If done manually it takes several weeks of detailed observation of utilization data, network traffic, IO transactions, and then mapping to equivalent templates in the cloud after searching through the hundreds of templates in the cloud catalogs. While cloud vendors offer calculators, they model one VM at a time, not a complete environment that can contain hundreds of heterogeneous VMs. Also, they do not discover on-premise environments or do right-sizing. Migration/provisioning tools either do not do any cost analysis, or offer cost ball-parks for pre-existing application templates.
The right way to perform cost modeling is with a tool that
– discovers the on-premise environment,
– models costs for VMs in bulk,
– retains the on-premise cost-performance tuning to carry over to the cloud,
– suggests advanced cloud services such as load balancers, networking services, OS versions, etc.,
– stays currents with latest pricing and deals to show cost saving recommendations across all the mega clouds, and
– performs all these steps in minutes
We did a review of over 5000 VMs modeled in our tool in the last six months and distilled some interesting factoids. Click here to read them.