Andy Jassy, SVP at Amazon, last week shared his thoughts on the private cloud
Last week Gathering Clouds favorite David Linthicum published an article on Andy Jassy’s (SVP at Amazon) public push back on private cloud. Perhaps we are meant to understand this as a jab toward Rackspace, but it brings up an interesting conversation about the differences, and distinct benefits, of public and private clouds.
One point that Jassy makes is that private clouds are not as cost effective as public clouds. This is simply not true. Public clouds are effective tools when used properly, but at high scale and high volume, public clouds become far more expensive.
Let’s take a case example:
If you need 1000 servers for 2 hours, private cloud will not win on price compared to public cloud. However, if you need 50 servers for 720 hours per month, private cloud has the better value proposition since the price for usage is both more distributed over a longer period and more predictable, since private cloud avoids the variable costs associated with public cloud.
Public cloud helps you get to your incremental cost breakdown through its usage. However, once you reach a certain sustained point in terms of how your run rate has grown, it makes a lot of sense, from both usage and cost perspectives, to consolidate to private cloud.
There are other considerations when it comes to the public vs. private cloud decision (see some of our own here). A chief concern is security. Public cloud is inherently less secure than a private, dedicated cloud infrastructure. This is because public cloud resources are shared. Hypervisors can be attacked to exploit memory. Your resources in one location can be affected by a DDOS of a client on the same system. The DDOS point, for example, could have a major impact on a company’s public cloud infrastructure depending on where it’s situated. As an Amazon Web Services (AWS) client, you have no control over where your services are situated so it becomes nearly impossible to move your infrastructure after the fact or even during that event.
Failure is also a key part of the public cloud by design. You can lose entire Availability Zones in the blink of an eye, and if you have planned correctly, this is not a major problem. While you should build with failure tolerance in mind, involving Disaster Recovery (DR) plans and redundancy into your considerations, not everyone can afford to do that. In many cases, it may suit a business much better to have a secure, dedicated environment that is inherently more reliable.
While the level of impact or concern that any of these issues should have for a client, are completely case by case, these are real issues that users of a public cloud need to understand and appreciate in order to make the best decisions where their infrastructure is concerned.
Both versions of cloud need to coexist because they appeal to complementary, but fundamentally different business conditions. Often, the best path for a business is a combination of both, constituting a hybrid cloud approach (at least in one definition). Through this configuration, a business can tap the benefits of a reliable, cost-predictive cloud model while accessing the pay-as-you-go benefits of the public cloud.
Jassy may have motivations for his statement separate from the actual functionality and benefits of the public cloud. Does AWS feel threatened by Rackspace? Probably not, but the statement is important in how we can evaluate what cloud actually means to a business based on the real conditions that govern its usage.
Let us know your thoughts on this article. Follow us at @cloudgathering on Twitter.
By Jake Gardner