A Cloud Solid Enough to Stand On

The City of LA's decision to replace its e-mail system with Gmail can serve as a rubric for what makes a cloud offering desirable to organizations.

The promise of cloud computing is, for most organizations, only a promise right now. Where's the data that organizations need to make decisions about when cloud-based services make financial and technical sense?

Vendors are coy with details about how and how many customers are actually using their services. They don't release real data, just Muzak from customers who have been compensated to lend their names to carefully manicured case studies.

Which brings us to Hollywood, or to be more precise, the City of Los Angeles. In a highly publicized competition between Microsoft and Google to replace a tottering Novell GroupWise e-mail system, LA last fall chose Gmail. Because the City of LA is a public organization, the details of this decision are freely available and serve as a valuable resource for anyone who wants real-world cloud computing data.

Some of the major calculations in LA's analysis involve the key issues that come up in any discussion of cloud computing: the reliability of cloud systems; where the savings, if any, really come from; and the security of data in the cloud.

Reliability. Google's e-mail system is expected to deliver better reliability than the city's on-premises system, which delivered about 99.89 percent uptime over the last year, compared with Google's 99.97 percent uptime in the same period. The difference, to spare you the calculations, is about five hours of downtime over the year.

Savings. The main savings will come from staffing reductions and reallocations. The 16 staff members who maintain and administer more than 60 e-mail servers for 30,000 users cost the City about $1.5 million in fully loaded salary costs. The City may also reduce its annual Microsoft bill (currently about $1.6 million a year) if it can wean many of them off Office and onto Google Apps.

Security. The LA Police Department has concerns about confidentiality and data security. Google is building a GovCloud, a segregated data environment that is physically and logically separate from other clouds, encrypted, located exclusively in the continental United States, striped across numerous servers, and accessible to only a small set of Google administrators. That got LAPD onside.

While I've been talking mostly about Google's cloud here, I'm not suggesting that only Google could deliver this package. The same advantages could be realized on other vendors' clouds, such as offerings from Microsoft and Amazon. Yes, there are some sweeteners from Google here, such as a $10 per-user rebate that the City gets if users from other California governments piggyback on LA's contract and adopt the same solution, but any vendor could have offered that. (The City claims that it has more than 20 serious inquiries from other California state and local agencies who are interested in leveraging its contract.)

What are the implications for channel partners? Take Computer Sciences Corp., which will be paid just under $900,000 for first-year implementation services. That works out to about $30 per user. That's slightly less than Google itself will get per year (I calculate that at about $40 a year), and it does raise a yellow flag for partners.

A 1-1 ratio between partner services and software/vendor revenue is low by historical standards -- deployment and integration revenues typically run five to 10 times the cost of software, for example. Microsoft partners implementing Microsoft Online Services report a multiplier of about three times the first year's subscription costs, so cloud computing does not look like a huge revenue opportunity for the channel in the long run. One factor to consider: e-mail implementations may be simpler than more complex portal assignments that could follow e-mail into the cloud and generate much more partner revenue.

In spite of these modest margins, with so many customers cutting back and making do with what is already in place, most partners aren't turning down a lot of integration and deployment work. LA had 15 bidders for its cloud-computing project; partners who can get in front of this wave stand a good chance of landing what work there is in what could be years of migration to the cloud.

About the Author

Paul DeGroot is principle consultant with Pica Communications, which provides consulting services for customers with complex Microsoft licensing issues.