Transparency In The Cloud, Part II: 3 Questions to Ask Every SaaS Provider
In Part I, we recognized that end-users of B2B software are in a unique position to drive the way internet-based services are delivered. This is important, because the data shows that most of us are going to at least start to transform the way we consume software in our businesses within the next couple of years.
We all know that answers lead to knowledge, and knowledge leads to power (Your first Jedi allusion of the week!). So how do we, as end-users, get the upper hand when it comes to software-as-a-service? We do this by asking the proper questions. To get you started, here are the first three questions you should ask every SaaS provider you encounter:
Question #1: How does your service scale?
Historically, software companies needed only worry about how well their software could handle the load within the walls of the largest customer they had. For most deals, they could get away with telling us that the software is used by a token Fortune 500 company that is 4 gazillion times bigger than us. We’d extrapolate from numbers well above our heads and inevitably say “Well, if it works for them, it must work for smaller companies like mine.” On the contrary, in the software as a service world, the logic often works in reverse. Remember that our software vendors are now responsible for mustering the ping, power, and pipe to provide service to all of their customers as one large entity. In the aforementioned scenario, that makes most of us small fish in a big pond. If a service provider doesn’t have a viable approach to scalability, which means more than just “throwing servers at it,” then they are going to experience something called “buckling.” When a service provider experiences buckling, and their service slows to a crawl, or stops completely, it makes the fact that they just landed another 10,000 seat Fortune 500 deal not so impressive to the rest of us.
Ask them how they scale. Do they provision a new server for every customer? (if they do, say “welcome to 1999,” and then run.) Do they provision a new VM for every customer? If so, how many VMs of their particular software footprint can fit on a single physical server in their datacenter? Have they built scaling mechanics into their application or are they relying purely on infrastructure growth? All of these things play into that “buckling” thing I mentioned, and often times you can gauge if and when a provider will buckle simply by knowing the answers to these scalability questions. For more on this, see economies of scale.
Question #2: How do you release new versions of your service?
Releasing software on CDs is safe. It’s safe because there’s an entire protocol around the release and there’s time between when the software is ready and when the bulk of customers will install it. That means there’s ample time to test the software and collect feedback, and there’s even time to fix mistakes after the software has been released. In the world of online services, there are no “take backs.” When a service provider updates their services, chances are we will see these updates the very next time we click a button in our browser or refresh the page. That gives them very little room for error. To circumvent this situation, and make life a little more familiar, some service providers declare scheduled upgrades. They carve out little windows of time to make the service unavailable to anyone but themselves, and figure the fact that their customers are aware of this impending outage makes it ok. Calling timeout gives them the opportunity to deploy the update and test it in the privacy of their own datacenters, just like they used to before releasing a new CD. The problem here is that their timing will never be your timing, and although it may not happen often, eventually there will be a scheduled update that conflicts with your need to access the service. The good service providers will find a way to upgrade you without downtime. Find one of them.
Question #3: How do you plan for disaster recovery?
Even the mighty Google is not impervious to the whims of server ghosts. Just ask the 40,000 or so Gmail users whose inboxes recently went *poof* in the night. But that’s Gmail, a free consumer-focused email service. Google rightfully offers little to no guarantee that this won’t happen to you or me tomorrow. This might be acceptable for our personal email accounts (what do you expect for free?), but would this type of “c’est la vie” approach to your business’s data fly with your CEO? Not mine. Therefore, it’s imperative that you know of your service provider’s disaster recovery and business continuity plans. Disaster recovery means a broad spectrum of things, so there are a lot of questions to ask. Start with a couple: If data is lost, how quickly can it be restored? Can another customer potentially do something that would result in an outage or data loss for me? If half of your datacenter falls into a spontaneous sink hole, how many customers would go with it? Hint on that last one, if the answer is more than zero, chances are their plan is not worth your dime.
There’s a lot more to the disaster recovery discussion, but I will leave it here for now, lest I spark a conversation that will have us exercising the disaster recovery plan for this very blog.
In Part III of Transparency In The Cloud, we’ll continue our list of important questions to ask all SaaS providers.