Design Criteria Defaults: SaaS

Yesterday I wrote about vendors making mobility a primary design criteria when developing collaboration platform interfaces, today I want to focus on the SaaS.

Still in it's 20o9 rage, the Cloud continues to pose issues for customers and vendors alike. Since late 2006 we've seen traditional software vendors, including IBM, Microsoft, Oracle, and Cisco, throw their hat and products into the Cloud and offering SaaS-based delivery of many of their popular solutions. Spurred on by Google's ambitions in the the enterprise market, each vendor has come up with their own approach to SaaS and the Cloud. Some have made it out of beta to deliver ready for prime time offerings (Microsoft Online, LotusLive), some have pulled back to only providing private offerings (Oracle), others are emerging through building on top of consumer-based acquisitions (IBM Lotus and Cisco), and others are working on configuring the right architecture (Cicso).

Bottom line, updating existing products and systems and building a hosting operation is not simple. It's not easy to transform an installed, on-premise system - like e-mail or collaboration - into a multi-tenant, scalable, and secure SaaS offering. The migration of back-end servers to support larger loads and parsing out multiple domains can take time. Once that's in place modifying other supporting services like Directory or Search to support the complex security and permissions also takes time. Finally, designing the presentation layer to provide secure functionality also requires a change in attitude and development assumptions.

As with mobility, vendors need to consider hosting capabilities as an primary design criteria for all new system software. Microsoft has done well in coming closer to this goal in its SharePoint and Office 2010 designs, but still some of their forthcoming BI servers are still treating the hosting part as a follow-on job. IBM Lotus on the other hand appears to be approaching things differently, developing on-premise solutions separately from it's LotusLive offerings, many of which consist of acquired hosted products like Outblaze's e-mail or the Unyte hosted meetings.

It's not a foregone conclusion that all on-premise platforms need to be retro-fitted or upgraded for SaaS delivery. Designing for hosting will likely help in the integration department and make hybrid on-premise/Cloud delivery easier for both the vendor and the customer.

Apples to Kumquats: Comparing Cloud Service Offerings

I recently read through a new Google Apps Premier Edition (GAPE) Partner Guide called 7 Keys to Comparing Google Apps Premier Edition with Microsoft Exchange Server 2007. Discrepancies and misdemeanors aside, the biggest problem I have with the piece is that it's the wrong comparison if you want to sell Google Apps to customers. The piece is great collateral to demonstrate the differences between on-premises and hosted services but it does not say why way GAPE is say better than Microsoft Exchange Online.

To be fair, and not to single out Google, I've found that the collateral from major cloud computing vendors (including IBM and Microsoft) tend to do a pretty good job at presenting arguments in favor of cloud-based services versus on-premises deployments, but they tend to confuse the comparison as justification that their particular services are better than another vendor's services. Wouldn't the Google piece have more value if it were called 7 Keys to Comparing Google Apps Premier edition with Microsoft Exchange Online?

Of course cloud computing is the hot new item and currently represents a huge new promised land for enterprise computing; what if you could have all of the benefits of email without all the fuss? It's like dating without commitments. In reality there are commitments and right now vendors see huge opportunities as long as they can get the huge economies of scale that make it all pay off. It's big business.

Last week I blogged about a recent McKinsey report that shoots some holes into the cloud argument. My best advice to any customer is to consider these delivery models carefully and make strategic decisions according to thier unique business, technical, and security requirements. The real comparison is two-fold; is cloud computing a good choice for you, and which cloud offering will meet the needs of your firm?

Thinning Clouds? Report Questions Value Of Cloud Computing - Software - IT Channel News by CRN

Follow the link in the article to the McKinsey report, it's a highly worthwhile read. I think the last comment sums of the net effect of McKinsey's observations and how nascent the market is, especially for cloud services:

But IBM said it is seeing many customers adopt a mix of public and private cloud models to reduce the cost of supporting specific applications such as business resiliency, information protection and collaboration services.

Thinning Clouds? Report Questions Value Of Cloud Computing - Software - IT Channel News by CRN

Cloud Services: What do you want?

I ran across this IDC chart on what customers want with cloud computing service providers. The survey is from August 2008 so it's pretty current.

Overall not so surprising that cost and performance are at the top of the list. I was a bit surprised that an established provider fell into a lower percentile. I guess that's mitigated by the #4 response, the ability to "move offerings back on-premise"; assuming that means being able to move data back on-premise. I would have figured that being an established provider would be more important.

A recent Information Week article on the closure of Goghead (a low-cost application platform hoster) highlights the risks of cloud computing, especially when the provider goes out of business. Still I guess if you're joining the cloud frontier you'd expect some risk. Even if it's a "large, established" provider there still are risks in the "how do I get my stuff back" area. As IW the article points out:

Choosing a stable vendor can reduce some of the risks of cloud computing. But even Google acknowledges that the migration path off the Google App Engine, should its customers want one, needs some work. "This is an issue, and we know it's not as easy as it should be," said product manager Pete Koomen.