PaaS, more than SaaS
In May 2007 I attended Salesforce.com developer conference where Salesforce SOA was announced as an add-on to the Apex platform. Industry analysts focusing on the area of SaaS considered the move as significant, but not surprising (see Phil Wainewright's take on the new in May 2007 as an example).
The game is moving on. PaaS, or "Platform as a Service" could be the acronym that defines a new web-oriented model where more than just specific vertical "services" are delivered as SaaS (e.g CRM, ERP, etc).
PaaS speaks to a more generalized services platform concept. If the "web as a platform" is the notion of multiple services in the cloud, then where does the "composition" happen? There are several Mashup design patterns to chose from. These patterns are along a continuum - from pure browser presentation mashups to client-side mashups to server-side services and data mashups (composite applications). If architecturally speaking you land in the composite applications pattern for delivering your SaaS apps, then this is where PaaS comes in.
This last week there is further evidence that Microsoft and Salesforce want to benefit from the opportunity in the power of PaaS. From SaaS out, PaaS in, by Angela Eager at CBRonline.com: Sub heading is:
"The Salesforce.com Summer '07 release is all about the platform, with software-as-a-service evolving to platform-as-a-service, a move that will antagonize and incentivize big league players Microsoft, SAP, and Oracle."
Angela's analysis includes a comparison to Microsoft's own forthcoming offering, including "Titan"
"Salesforce.com might be the first to deliver an on-demand platform but it certainly will not be the only one, and the on-demand specialist is starting to feel Microsoft's breath.
Releasing snippets of information about the forthcoming Titan offering at its partner conference last week, Microsoft chief Steve Ballmer said Titan referred to a CRM application-development platform on which developers would be able to build customized applications that could run on their servers or on Microsoft's own servers.
Although the Microsoft offering is not yet available, and even after launch it will take time for some of the planned functionality to be available such as running custom user applications within the Microsoft data center, it will be a direct challenge to Salesforce.com. With similar platform offerings, users will have a choice for the first time, which has the potential to cut into Salesforce.com's growth."
Maybe "Titan" is the news I was looking for last week, but we're seeing hints at more than just Dynamics APIs as SaaS from Microsoft. There is still no real hard news on this today, but watch out for posts by the likes of Phil Richardson, a Microsoft Program manager in the Dynamics team (and my ex-neighbor in Redmond) like this where Phil hints:
"Soon (can't say exactly when) I'll be starting to blog about Titan features....Once marketing gives the go-ahead you can expect to see boatloads of Titan content on this blog. I'll be focusing primarily on features which are targeted at VARs & ISVs - but I'll make sure to throw in some end user features."
And then Phil hints a little more this week...
"Some vendors have the 'heads in the sand' and refuse to talk services seriously (with no offering or a token offering). Other vendors only offer SaaS and trash talk those who aren't pure services players. There is no question that our industry is moving towards services and Microsoft's CRM has a strategy which is sensitive towards these industry changes. As we announced: we are using our Titan Project to create a Microsoft Hosted version, a Partner Hosted version and an On-Premise version. I use the term 'version' loosely as the final number of SKUs is up to marketing. I believe we are entering an time in the industry when customers will start moving away from self hosting 'non-unique' business functions. Anything which can be achieved by configuring on the shelf apps (like Microsoft CRM) will eventually move to hosted services (if I knew exactly when 'eventually' was I'd be a rich man). Our strategy understands that customers and partners will find themselves in varying degrees along the on-premise to hosted spectrum. Some will want everything 'in the cloud' and others will want everyone their own datacenters. Some might want to prototype in the cloud and move to on-premise etc etc. I believe vendors need to interpret these changing times appropriately for their business."
(hey Phil....trackbacks? comments? common dude!)
My take is that Phil hints to more generalized services platform concept from Microsoft. PaaS, more than SaaS.
Delivering on PaaS is hard though. Yesterday, Dan Farber's post Salesforce.com transitions to platform as a service highlighted one of the greatest challenges in this PaaS space - uptime at scale:
"Collins said that salesforce.com consistently provides above 99.9 percent availability. “Three nines is the best quality of service you can get today from any on demand player,” he said, claiming that the best deployments of Siebel and Oracle are closer to 96- to 97-percent available.
In any case, as salesforce.com scales up the enterprise and companies have more dependence than for just CRM on its platform, three nines, which doesn’t include times when the service is up but suffers performance problems, won’t be sufficient for customers who can’t afford 8.76 hours of downtime a year."
Uptime is just a small (but critical) component of the whole PaaS vision. Developer tools for PaaS is another kettle of fish (see Dana Gardner's post on Software Development and Deployment as a Service, or "SDDS" - on this topic). It's how Bungee Labs sees the future. But that's for another post.