Creating a Hybrid Management SLA Dashboard
Regarding management of the SLA of a hybrid cloud environment, think in terms of a dashboard that allows you to have visibility into the combination of services and then the ability to detect which part of your hybrid environment is causing the problems.
Here is an example that might help you think about this issue: Imagine that you are in charge of managing a sales force. You have a direct sales force, a group of contract sales people, and a distributor that sell your products. You look at the overall performance of the entire hybrid sales organization, and you notice that sales are off by 25 percent from the prior month. What does this mean? Is everyone having problems? Instead of panicking, you take a look at each group. It turns out that the distributor was acquired by another company and the team that had been selling a lot of your products was reassigned to different accounts. In fact, when you take out the performance of the distributor, it’s clear that the rest of the sales organizations are outperforming previous months by a wide margin.
When downtime isn’t an outage (by SLA terms)
If an SLA states that a cloud vendor will be up for 99.999 percent of the time (trailing 365 days), you likely expect them to be down for around 5 minutes over the course of a year (5.256 minutes to be exact). When viewing the many 9s, outages might be far from your mind, but a close reading of your SLA is important in order to understand what an “outage” is under the terms of the SLA. For example, a major cloud vendor was partially down for four days but managed to not breach its SLA. Through an act of goodwill, affected customers were compensated, but they did not have recourse under the 99+ percent uptime SLA.
Technical staff should be brought in to consult on SLAs. Simply reviewing these documents on their face and nailing down the terms and conditions to make sure you’re compensated when services are not delivered is not sufficient. You must closely read potential SLAs, understand their terms, and know how the resources will be used in your organization. Technical staff should be consulted on vulnerabilities and choke points, and the organization must understand its risk profile as it ventures forward.
Here is a — certainly not exhaustive — list of some occurrences that may lead to downtime without recourse under the SLA:
Maintenance (both scheduled and emergency): Think about what kind of notice you expect to receive and whether your cloud provider gives such notice. What is maintenance and more importantly, what is emergency maintenance?
Attacks by third parties: Whether the third party is maliciously targeting you, the infrastructure, or another user on the cloud, the SLA may not cover such downtimes. What sort of security does the cloud provider have in effect?
Tip: Also you might want to consider your cloud provider’s other customers and whether they’re targets for hackers.
Unavailability due to third-party system failures or services not under the vendor’s control: What third parties does your cloud supplier rely on? Do they have a strong partnership, or are they simply employing fly-by-night companies?
Force majeure: These include acts of war, earthquakes, floods, riots, labor strikes, and so on. Think about the actual data center where your cloud services are located. Is it in an area prone to extreme weather or unrest?
What if you take the same approach when you analyze your hybrid cloud environment? We think that is exactly what you will need to do. Of course, some requirements have to be in place; for example, you will need
A catalog that keeps track of performance statistics from each cloud environment
An API that allows those statistics to be visible from the dashboard
Metadata that interprets what each performance measure means from one environment to the next
Tools that allow you to drill down on each element of your hybrid environment
There are indeed tools from many cloud management vendors that are helpful. However, there is still work to do before there are solutions that allow an IT organization to have a comprehensive way to have an overall view of the relationships of all the cloud services a company might be dependent on.