I was catching up on RSS today and read a piece by Alex Bainbridge on Tnooz about the Top Eight Travel Industry Battles.
A good article overall, but battle #3 jumped out at me and I guess I’m ready to throw down a little.
First of all, I don’t disagree that "The Cloud" is susceptible to some of the same problems as a traditional data center, especially the most noticeable of incidents, unplanned outages.
But that is a far cry from saying that Cloud is just a data center, as Bainbridge implies.
There is already a lot of FUD and confusion on a topic and many of the likely readers don’t know any better and can be easily swayed by a comment from a respected technologist such as Bainbridge.
So let me get this off my chest:
- The Cloud is not just another data center. It’s not hosting or the same thing as "time-sharing" either.
- The Cloud can fail. It is not magical. It is even more vulnerable to certain events, like a cascading failure caused by another application in the same node, that a segregated data center would not be. On the other hand, the leading cloud providers (Amazon, Google, Rackspace, Microsoft, etc.) put resources into creating a resilient architecture and monitoring that most businesses could never dream of, making Cloud significantly more resilient, reliable and secure than would be delivered by most standalone data centers supported by an internal IT staff.
- Make no mistake, application resiliency and security is an outcome of good architecture and design. Applications can fail equally in the Cloud, a data center or on your mobile phone if they are conceived and developed poorly.
- Yes some data centers use many of the principles used in the Cloud like virtualization and automatic provisioning. But don’t assume that all do. Infrastructure architecture is critically important and not easy. Do you think your IT staff is smarter than Werner Vogels and his crew?
But as we look at the battle between Cloud and the traditional data center, consider these points:
1. CAPEX v. OPEX:
If you build a data center, be prepared to write a big check and take time to procure & install the hardware. Cloud is pay as you go.
This makes a big difference in cash flow, funding requirements (if you’re a start-up/early-stage company) and how quickly you can react to business growth.
On a related note, when you buy the equipment for a data center, you need to buy based on peak load.
One of the key benefits of cloud is elasticity… you don’t buy capacity you pay for resources that you use, and that varies quite a bit whether it’s August 3rd or Black Friday.
With the Cloud, computing resources are automatically provisioned to meet demand, but scale back (as does the cost) when demand ebbs. Is this important for your business?
3. Performance and latency
Is your business global? If you have customers in different parts of the world and the performance of your application is critical (e.g. impact of return of search results on customer conversion rates), latency is important.
If you force a customer in Singapore hit a server in New Jersey (because that’s where your data center is), the latency in response and experienced performance by the customer suffers.
Alternatively, if you used the Cloud, you can have instances in different regions of the world that provide superior performance than a single data center could provide (of course you can build replicated data centers over the world, but that’s certainly not cheap).
Further, as the world turns, with the cloud you can ‘follow the sun’ and vary the compute capacity across the globe to match customer demand. That can help normalize the amount of computing resources you pay for and help keep costs… in addition to performance… more predictable.
4. Disaster recovery:
Another great benefit to Cloud over traditional data centers is disaster recovery. Imagine you had a data center in Lower Manhattan a few weeks ago.
Getting systems back online when they are under two feet of water is hard and doesn’t go especially fast (too soon?). With the Cloud, shifting the instance from one affected location to another is relatively fast and simple, and can be done in anticipation of catastrophic problems.
So yes, the Cloud is not perfect, but it does offer some significant advantages. Whether they are right for your company is an open question. And it’s not all technical questions.
Business needs must be where the discussion begins. And IT owners must be honest with their evaluations of the technical advantages, but also of their own capabilities and personal agendas.
So as you can see, the decision of whether to adopt Cloud computing can be difficult to assess, but shouldn’t be further complicated based on a misunderstanding of what it really is.
NB:Cloud tablet image via Shutterstock.