Cloud Computing & SLAs

16 Nov

A Service-Level Agreement (SLA) is a document describing the level of service expected by a customer from a supplier, laying out the metrics by which that service is measured, and the remedies or penalties, if any, should the agreed-upon levels not be achieved. Usually, SLAs are between companies and external suppliers, but they may also be between two departments within a company.

The five most important elements of a cloud SLA, as outlined by analyst Robert Desisto of Gartner Inc., are given below. I have added a bit more detail to each element so as to give a clearer understanding of what exactly they represent. (Gartner Inc. delivers technology research to global technology business leaders to make informed decisions on key initiatives).

General availability

You generally want between 99.7% and 99.999% uptime; anything mission-critical has to be 99.9%. This can be measured by time slot. For example, requiring 99.7% availability between the hours of 8am and 6pm, and more, or less availability specified during other times.

E-commerce operations typically have extremely aggressive SLAs at all times. For example, 99.999% uptime is not an uncommon requirement for a site that generates millions of dollars an hour. 99.999% works out to about 5 and a quarter minutes of downtime per year which can actually be too long for some businesses’ preference.

Measurement period

This determines how the vendor quantifies its uptime performance. Typically, the measurement period will be a month or a quarter. On the one hand, a longer measurement period will give the vendor more opportunity to make up for bad performance, while on the other hand, a shorter measurement period gives the vendor a “fresh start” more often. Thus, it is important taking this into account when deciding the measurement period.

In addition, the SLA should define whether the measurement period includes times when the vendor is shut down due to severe weather, acts of war, failure of its suppliers, etc. Furthermore, does the measurement period include periods when the demand for services exceeds the assumed levels?

Incident notification

This is where the vendor will notify you of a problem, and not rely on you to report the incident. The SLA might also require the supplier to conduct a root-cause analysis of service level failures and report the results to the customer.

Application of penalties

What is the financial remedy applied to you if the SLA is violated? Is it a percentage of fees or a credit on the following year’s invoice? Or will they prorate your bill? The SLA can provide increased certainty by defining particular events that, without argument, allow termination for cause. For example, a minimum service level can be set at the level that allows termination. In addition, a certain amount of accumulated service level failures can be defined that allow termination for cause. The effect is to give the customer a clear exit right for substandard performance.

What constitutes downtime?

This could be anything from 5 minutes to over 5 hours. Different vendors will have different ideas of what constitutes downtime. Below is an example of part of a SLA for OpSource Cloud Hosting. It illustrates the percentage of the monthly fee that is given as credit in relation to how much downtime occurs. For example, anything over 2 hours and under 3 hours downtime over the course of a month will result in a service credit of 15% of the monthly fee.

OpSource Cloud Hosting SLA Example

“The importance of service-level agreements is overblown when it comes to cloud services” is the shared message of executives from top cloud service providers such as Microsoft, and What do you think? How important are SLAs when it comes to cloud service agreements?







5 Responses to “Cloud Computing & SLAs”

  1. billynomates2012 November 16, 2012 at 6:35 pm #

    Is an SLA another version of KPI or is it used as a KPI for a business

  2. steepletoes November 16, 2012 at 7:14 pm #

    Not exactly, but the two terms are very closely related. A KPI can be thought of as a measurement that tells management the precise state of affairs at any given time, i.e. a measure of performance. The SLA is the ideal state of those measurements.

    For example, uptime is a general KPI for a cloud service, while 99.7% uptime may be the SLA, i.e. the ideal state of affairs. When the KPI deviates from the desired SLA i.e. uptime deviates from 99.7%, it must send some sort of a message to the management highlighting that management interference is required.

    To summarize, the KPI measures the performance level of an element (e.g. uptime), while the SLA defines the ideal, or specified level at which the element should be performing.

    Hope that answers your question ‘billynomates2012’.

  3. billynomates2012 November 16, 2012 at 9:44 pm #

    Thanks for your response. That was a extremely informative response and has cleared up any confusion i had between the two

  4. agblogail November 20, 2012 at 10:43 am #

    After recently reading Michael Earls article “the risks of outsourcing”, he spoke at lengths about SLA’s and their importance, but does not define them, The article is about outsourcing SLA’s and would be very similar to Cloud SLA’s, if not identical. Thank you explaining SLA’s in detail as Earls article will make more sense as a result.

  5. steepletoes November 20, 2012 at 6:25 pm #

    Thanks for your comment ‘agblogail’. Glad I could help. In relation to Michael Earls’ article, I would agree with his view that SLAs are of great importance, especially for the customer. I think the likes of Microsoft, and are playing down the importance of SLAs simply because they are the vendor, and any way in which they can lessen their responsibility amid a breach of a SLA is a route they will explore.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: