Your Cloud Availability: 98%, 99.99% or 99.9999%? CloudEndure Gets Inside

I recently had a discussion with a few very interesting cloud professionals from CloudEndure about the changes that the cloud and AWS in particular, bring to the table in terms of business continuity and high availability. In one of my discussions with Ofer Gadish, the company’s CEO, we exchanged information regarding the current state of the market, including tools, terms and relevant methodologies for real cloud users. We both agreed that in this market stage business continuity in the cloud is a very interesting subject since the market is still young and the traditional tools and perceptions don’t apply.

While we saw eye-to-eye on many topics, many questions arose in regards to specifics on the topic of cloud market maturity. In order to answer these questions and expand our cloud communities’ knowledge, CloudEndure decided to run a survey to learn how the cloud has changed DR, Backup and High Availability.  And we’ll share the results with you.

 

What’s Downtime?

We discussed my belief that the nines (99.99…%) are still marketing than the real deal today. The market and tools used to measure your uptime immature or do not really exist. The concept of availability in the cloud is determined by the level of responsibility and liability that the vendors have for their customers. These notions include the ability to monitor, proactively fix and maintain continuous communication with users, giving them clear and genuine visibility into what exactly is going on and when the system is expected to return to normal. It is also necessary, here, to define the concept of downtime. At a very basic level, it is when the system is not available. However, the more precise answer depends on the criticality of specific features and components of an application or service.

On that note, we are interested in hearing your point of view. First and foremost, it is essential to ask, what do you consider “downtime”? How do you approach the matter of downtime with your customers? How do you compensate, if at all? Maybe even before discussing compensation, let us ask how or if you measure downtime? If so, are you able to calculate your own availability over the past week, month, or year?  With your help, we can gain even greater insights on the topic. 

RTO and RPO – Nothing Really Changes


Definitions first:

Recovery Time Objective (RTO) is the duration of time and a service level within which a business process must be restored after a disaster (or disruption) in order to avoid unacceptable consequences associated with a break in business continuity. Wikipedia

Recovery Point Objective (RPO) It is the maximum tolerable period in which data might be lost from an IT service due to a major incident Wikipedia


The cloud changes the ways in which you achieve optimal levels of RPO and RTO. There are much more creative ways now-a-days to deliver robust solutions at a very low cost.  Service degrades and downtimes can occur as a result of a bug, application scalability limitations, human error or even your cloud operator’s availability. In any case you, as the service provider, have the responsibility to get it up and running again. Keeping the application, and its actual business values, in mind, you can address both RPO and RTO and make sure that you define and measure these metrics.

Do you know your targeted RTO and RPO goals? Are you able to meet these goals?

DR Cost Impacts and ROI

When you’re an online retailer, every moment that your website-relevant functionalities are down, you lose money. Your business continuity should be prefect including accurate measuring capabilities for your site’s uptime performance as well as ensuring there are no latencies or degrades that directly affect your business.

 “…breaks down to about $120,000 per minute. Multiply that by the 40 minutes Amazon went down and you get around $4.8 million in lost dollars.” Geekwire release following Amazon downtime on Aug-2013

A DR policy and rate is based on the required SLA; however good DR deployment on the cloud should be based on the central, pay-per-use cloud principle. You can actually create a spare cloud replica without the traditional upkeep costs if it is not utilized. Today, cloud users can avoid traditional costs by using an “On-demand DR” inline with changing their DR strategy and architecture.

Do you know your monthly DR costs? What’s your estimated cost of downtime?

Final words

Over the past year, the “average enterprise” began its journey into the cloud, yet still views the cloud as an on-premise extension used as a means to generate replications and deal with less significant data. I believe that the sensible next step for companies that have already made this strategic move should be to move their production and more critical substance to the cloud. This would raise the bar and present the greater challenge of providing proven robust delivery of online services. Little by little, the cloud is making its presence and advantages known in a number of aspects of our daily lives. It’s up to us to embrace it.

  • http://www.itoctopus.com/ itoctopus

    When I read this post, I immediately remembered that every hosting company (cloud or not) that our clients use claim 100% availability (or 100% SLA), some are humble to claim 99.99999% availability.

    In my opinion, they are all making false claims, and they only honor their SLA when you call them the moment the problem happens, and nearly every day there is a problem: Problem on their network, problem on the VPS account that is hosted on the same server that your VPS account is hosted on, problem on the cloud…

    Even a slow website (for reasons unrelated to the website) is, in my opinion, downtime, and can lead to loss of sales, or indirect loss of revenue. Unfortunately, not a single company out there has an honest tool to monitor the availability of each account, and we have to rely on 3rd party, external tools to monitor the availability of our websites.

    The majority of hosting companies out there don’t take their clients seriously – and they think that a few minutes of downtime/slowdown every day won’t harm anyone. Maybe that mentality will change at one point.

© 2014 IOD