The AWS re:Invent 2013 experience
A comprehensive collection of articles, videos and more, hand-picked by our editors
Here's a way for IT teams to win end users' hearts: Create a cloud-based self-service portal where they can provision their own IT resources.
But there's a caveat: The level of affection ultimately depends on just how well the portal works and how easy it is to use.
Users don't want to know what's secure and what's not. They know that they need a service, and that's what you want to give them.
engineer, RightScale Inc.
Rishi Vaish and Ryan Geyer can sum up their advice for developing an effective cloud self-service portal in just two main points.
First, keep things extremely simple, at least from the user's perspective, said Vaish, vice president of products for Right Scale Inc., a Santa Barbara, Calif.-based provider of products for multicloud management. Users should see just the tip of the iceberg -- an easy-to-use interface allowing them to provision their own IT resources in the cloud.
Geyer, a RightScale engineer, agreed. "They don't need to know what's secure and what's not," he said. "They know that they need a service -- and that's what you want to give them."
Second, but equally important: Address five key issues before launching the cloud self-service portal, Vaish said. Those issues are: consistency, repeatability, governance, cost management and choice.
Vaish and Geyer shared their recommendations during a session at the Amazon Web Services (AWS) re: Invent 2013 conference, which drew more than 9,000 business and IT pros to Las Vegas. RightScale is an AWS partner.
Five issues to address
Vaish outlined the five areas of concerns in more detail.
1. Consistency. As a shared resource, a cloud environment is particularly prone to reliability problems, he noted. Ensuring consistent service quality is essential for the cloud self-service portal's success.
2. Repeatability. "You want to make sure that each time you set things up, they're the same" across all clouds, Vaish said.
In addition, many challenges related to both reliability and repeatability can be solved through automation, Vaish said. As an example, he cited the need for updates and patches: "You're not going to go in and patch multiple servers. You need a tool that does that." Monitoring, alerts, auto-scaling and other functions can be automated as well.
3. Governance. "There are two dimensions to this," Vaish said. "One is just securing your resources in the cloud." The second is being able to audit past activity -- or, as Vaish put it, roll back to see "who did what when."
4. Cost management. Addressing the cost factor can be especially challenging, Vaish warned. "It is different than what you did in the past," he said. "This is much more variable." That's due to the cloud's ability to scale usage -- and costs -- up and down as needed.
IT teams must decide how to charge the various lines of business for the resources users consume, Vaish said. He recommended the following steps for using cloud analytics to better manage costs:
- Visualize all IT costs
- Forecast future expenditures
- Optimize expenditures and reduce waste
5. Choice. If the portal's menu is too large, users may be overwhelmed and only use resources they know -- or steer clear of the portal entirely. For that reason, Vaish recommended zeroing in a specific portfolio of key resources. "Giving them a more narrow choice is something you can do to help the lines of business," he said.
Organizations that successfully address these five issues are the ones most likely to wind up with self-service portals that work, Vaish and Geyer said.
As an example of success, they cited a cloud self-service portal (or, as Vaish put it, "an IT vending machine") that their seven-year-old company built for Eli Lilly and Company, the Indianapolis-based pharmaceutical giant. Among other requirements, Lilly wanted to deploy the portal in select IT resource pools based on jobs and in-house groups, and use extensive consumption metering.
RightScale developed a portal consisting of its own product on top of AWS, stocking resources that Lilly's various lines of business wanted employees to access. The result: Lilly has reduced average provisioning time from 36,000 minutes (about 600 hours) to about 30 minutes, and it now measure its portal costs in terms of cents spent per hour. Those are results that just about everyone can love.