Adding visibility to a Cloudy environment
Cloud operations are a black art. Cloudstatus.com gives some visibility into the vast systems.
When Hyperic launched CloudStatus.com, I think most people took a look and figured they would never need such a service. Then all of the sudden we saw multiple outages from Google and Amazon and CloudStatus.com became the best resource to figure out what was going on.
I asked Javier Soltero, CEO of Hyperic to provide some insight as to why the Cloud is as much of an operations as it is a deployment battle.
Guest post by Javier Soltero, CEO of Hyperic
Operations is one of the key open issues that will define Cloud computing's future.
The separation of Cloud offerings aroundmakes a lot of sense. Regardless of the use case, the idea that a business might choose the Cloud as a platform to build and consume applications because it inherently reduces or removes the operational burden is ridiculous.
The simple reason is that software, regardless of who is developing it, always fails. Those who refute that point haven't been around technology long enough or haven't paid attention to the fact that every single 'Cloud' has had outages recently.
Enterprise software consumers (the folks whose money most Cloud providers are looking to get) know better than to assume that any new platform (whether it's "the Cloud" or Linux or Java) is inherently management free. Because of this, I'm confident that until management technology (including everything from provisioning to monitoring) matures, the enterprise will still regard the Cloud as a place to do science experiments.
The big question is what role do Cloud providers have in providing management technology to their customers. All Cloud providers have management tooling they use for their own internal operations (a lot of them use Hyperic HQ!).
The problem is that consumers of those services are looking at the services from a completely different vantage point than their providers. Taking the "power utility" analogy often applied to Amazon.com as an example, you can imagine the difference between the tools used by folks like PG&E to manage the power grid versus the single meter and breaker panels you have in your own house.
Those meters at your house are there to provide the necessary 'local' management for a utility whose backend infrastructure you don't see, can't understand, and don't care to manage. Without those meters and breakers you wouldn't trust billing, service levels, or even be able to manage problems like broken appliances. To take this one (rather ridiculous) step further, imagine if every time you wanted to install a new light fixture you had to call up PG&E to tell them to shut off power to that section of your house so you don't fry yourself.
So, what are the Cloud's equivalent for the on-premise electric meter and circuit breakers? Many smart people are chasing to answer that question. Unfortunately, as Dave and others have identified, the Cloud does not have the level of standardization that the power grid does. Each Cloud runs at its own 'voltage' and has 'power sockets' that are shaped differently and that will make it very difficult to come up with a single solution. There's been a lot of discussion about the impact openness and standardization will have on the viability of the Cloud as a platform but not enough of that discussion has focused on where manageability fits in.
This will take some time to sort out. In the meantime, Hyperic has chosen to start by providing a transparent, free, third party mechanism to demonstrate the health and performance of the various Cloud offerings called CloudStatus.
In order for the Cloud to be trusted like the power grid is, there needs to be a place where both Cloud providers and consumers can see real-time information about how these complex environments are behaving from a customers' perspective.
CloudStatus is monitoring the various Clouds by both running inside and outside their respective environments, and displaying the performance and health information in a way that is simple enough for the customer to understand. The service was also designed to provide visibility into multiple Cloud environments regardless of the type of offering.
Ultimately, the goal is to give the consumers of the Cloud visibility while allowing providers to prove the reliability and performance benefits of the platforms they offer.
Javier Soltero, CEO, Hyperic
Disclosure: my company has a partnership with Hyperic. In the past we have also battled for Wii supremacy.