I wrote a post a week ago about lessons learned from the Dot-Com bust. In the 1990's, startups were a dime a dozen and VC money was flowing endlessly. Back then, startups were requiring huge sums of money just to get their infrastructure in place. Many of these companies could not afford the initial funding to build scalability into their infrastructure...a huge risk!
Now it's 2008 and VC money isn't as easy to come by, especially at the enormous amounts we saw in the Dot-Com days. The challenge many startups have is keeping the company small to control costs, but at the same time quickly bringing a product or service to market. These two goals can conflict. One approach is to limit your startup costs in your initial infrastructure build out by outsourcing your infrastructure needs to a PaaS (Platform as a Service) provider. This meets both objectives; control costs and quicker deployment. Let's discuss how.
On the surface, the cost of infrastructure in the cloud may sound very expensive. But let's not get stuck on the "sticker price". Instead, let's talk about total cost of ownership (TCO). If I am running a new startup with the requirements of hosting web based products or services that require 99% uptime, the startup costs are huge. Here is a short list of things I would have to shell out money for:
- Software licenses
- Load balancers
- Lease or buy data center floor space
- Business continuity and disaster recovery capabilities
- Data storage devices
- Operations and Systems Administration personnel
- Capital funds for the projects to put all of this infrastructure in place
Quicker to Market
In addition to all of these costs, the time it takes for a new team of people to assemble all of this infrastructure is not minimal. For startups who stick to standards and refrain from using proprietary technology solutions, they can have their infrastructure up and running in the cloud in just a few weeks or even days. The PaaS providers will say hours, but I am including the time it takes the company to validate and tweak the environment accordingly. Here are some of the time killers that can be avoided:
- Long vendor assessment initiatives for the various hardware and software vendors
- Multiple procurement processes for each vendor
- Multiple hardware and software installations
- Recruiting and hiring costs coupled with ongoing payroll costs
- Leasing or purchasing of real estate for datacenter(s)
- Time and costs for telecommunications and security efforts
One thing that I am not saying is that security doesn't matter. It matters more then ever when you outsource your data management. What I am saying is that you don't need to invest the time and money upfront to build out a highly secure infrastructure. Instead, you need to do a thorough vendor evaluation and choose the best provider to meet your infrastructure requirements. Then you have to manage your vendor and ensure that all of your agreed upon SLAs are being met.
Startups better suited for PaaS then established companies
Startups have a huge advantage over companies that have been around a while. It is much easier to start in the clouds then it is to move legacy systems and infrastructure into the clouds. I wrote a post on CIO.com last week about how SOA can get you to the clouds quicker. Startups who can see the benefits of cloud computing should take a service-oriented approach in the development of their products and services. The more abstract and loosely coupled their software is, the more agile and flexible they will become. It will also allow them to switch PaaS providers if they are not getting the level of service or support that they want. Another advantage is it can allow the startup to actually leverage two PaaS providers to eliminate the risk of downtime in case one provider fails. Having this option gives a startup the ability to load balance across PaaS vendors and send the majority of the traffic to the lower cost provider. I haven't seen this done but it is a great opportunity to use leverage on the vendors.
In summary, startups can reduce their upfront costs of infrastructure and staff by starting in the cloud. Long term (5 to 10 years), PaaS or utility computing will likely be the norm. So get a head start and launch in the clouds today. Focus on your core competency while your competition continues to spend millions to keep the lights on.