If your plan for cloud computing is to simply “move to the cloud”, it's time to rethink your plan. Success is not about “moving”; it's about changing to cloud computing. Everyone wants the business and technology advantage of cloud platform; the question is not about why, but how.
Don't underestimate the odds of a struggle in moving to the cloud platform. Nevertheless, there are some significant hidden benefits. With the right guidance, you can confidently make this bet: the cloud platform will deliver upside for your business. Here's an outline of the homework (with more details below) that we think you'll need to succeed:
- Start, but don't stop with TCO
- Using the cloud for app replacement parts can save big bucks
- DevOps practices can work before, during, and after migration
TCO is necessary, but not sufficient
Any cloud vendors worth their salt have a TCO calculator. Rackspace, Amazon, Google, VMware – all have done their homework to save your finance and IT staff a lot of spreadsheet headaches. It's true: comparing data center operating costs to the professional efficiency of the public cloud giants could put a spring in the step of any young CFO.
In any TCO analysis, there's an imaginary checklist with old costs on the left and new costs on the right. But there are some critical differences, starting with the expertise required to engineer a safe landing on the new infrastructure. In many cases, given enough time, your in-house IT professionals can learn those differences.
During that time, however, you can't count on them to do their day jobs. You'll have to keep both old and new infrastructures running and paid for at the same time.
That's expensive (and will make your CFO age faster). So is downtime, as people who haven't done it before need to learn by making mistakes. How expensive? Not long ago, Forrester Consulting wrote:
“less than 40% of firms meet or exceed goals related to both migration and run costs. … 58% of respondents claim the costs to run their infrastructure in the cloud were higher than estimated. … when asked what they would do differently in their next migration, the most common responses were to invest in tools, increase performance visibility, increase education of costs, risks and benefits, and spend more time on application assessment.”
Cloud there yet?
Contact us. Let's see how CloudGeometry can help.
To make the economics of costs subtraction from the left column to the right work, you'll need expert help. It's not only about getting it right the first time; it's about making sure that the transition happens with minimal disruption to the business. You probably want to do better than “minimal disruption” (which is why we think you can benefit from our expert help).
Hidden costs in application infrastructure
A key blind spot in TCO analysis has to do with understanding the economic impact of your applications. "Better wood behind the arrow" by adopting cloud infrastructure does help. But it's the apps that are the sharp-edge of your business arrowhead. There’s a great deal to gain from adopting cloud-native services as part of your application: databases, load balancing, distributed business process logic, security, compliance and more. The Amazon Web Services (AWS) product portfolio offers an incredibly rich set of options.
Why try to save on apps when hosting costs less than your datacenter? At its simplest level, the software licenses you pay for those application components today are far more costly than their cloud equivalents. Some of this has to do with classic (pre-cloud) licensing schemes, often premised on (and engineered for) the silicon-based CPU footprint of the data center machinery where those applications ran in the past. Merely moving them to a virtualized environment (e.g., MySQL hosted on EC2) leaves a lot of money on the table. Cheaper hardware-cost equivalents do nothing to cut the burden of human DBA overhead, for example.
There is real upside in shifting from a conventional RDBMS to Amazon's Aurora RDS suite. First, the data storage in virtualized SSDs is integrated by default with the database engine. There's no need to pick and choose where the data is stored vs. where the database runs. Automated backups are always enabled on Amazon Aurora DB Instances. Backups do not impact database performance. And consider this about performance: Aurora can give you up to a 5x benchmark speedup “by tightly integrating the database engine with an SSD-based virtualized storage layer purpose-built for database workloads, reducing writes to the storage system, minimizing lock contention and eliminating delays created by database process threads."
Your database does not run in a vacuum; it's the beating heart of your applications. None of the plethora AWS-native technologies are going to create efficiencies for every one of your apps in every situation. Making the most both of migration and the integration with your applications requires thoughtful planning and expertise, and that's where we can help. It's also the inflection point where your cloud migration goes from efficiency to competitive advantage, which is why you wanted to undertake cloud migration in the first place.
The cloud model for operations
It's common knowledge that DevOps and cloud computing go together; the hype seems to suggest that one requires the other. But the interaction between them is a little bit more nuanced: in fact, it's possible to execute cloud migration without taking on a DevOps model. More importantly, it's feasible to take on a DevOps Model without completing a cloud migration first.
DevOps and the cloud model proceed from the same motivation: create and deploy new features and infrastructure improvements faster, at a lower cost. In some ways, it's here that the notion of “migration” can most easily lead you astray.
You don't even have to finish cloud migration before you can begin to take advantage of cloud infrastructure. You could start with as simple a task as test automation. Cloud-hosted testing makes it straightforward to spin up and tear down an execution environment with every iteration of testing. It's a great way to get cloud going even before you nail down a complete cloud migration strategy. It contributes directly to the quality of the software that your business depends on, yet another way of "wading in" to the economics of cloud.
Here's one final example. Moving to the cloud operations model means that you're going to have an enormous amount of flexibility in your infrastructure. Moving faster also means that changes have faster impact. You'll need to monitor and manage the infrastructure to make sure that flexibility continues to deliver returns, with continuous oversight of the utilization, compliance, security etc. That means staying on top of costs; we've seen that disciplined, scheduled assessment and tuning of cloud utilization can regularly reap 20% savings.
A plan for a plan
There are many ways to introduce changes into your application estate and architecture, all depending on where the integration of the new cloud-enabled computing helps make your business more competitive.
Because we've been practicing the disciplines of continuous integration and cloud operations as our standard operating procedure, we can help you find where the adoption of cloud computing pays off, and in what sequence. We mentioned the example of databases, but you can use it for backups, for delivering new application functionality, integrating across other SaaS-based applications and so on.
We can show you how to choose where cloud computing practices will have the most significant impact first, and add new functions as your application needs mature. (We can even build a lot of those applications for you to help you move even faster).
Net net: operating on the cloud platform can deliver more bang for your buck in both time and money. The best way to judge the costs and benefits of a successful migration plan is in its flexibility. That's how the transition to cloud-based computing and applications ensures you lock in the gains, and that’s where we think we can help. The end game is just not getting to the cloud, but running on the cloud, by default. All the TCO and virtualization in the world doesn’t change the most basic assumption of cloud computing: time is money.