Advertisment

Want cloud, but outages put you off? This is for you

author-image
Deepa
Updated On
New Update

BANGALORE, INDIA: Cloud is here to stay, and so will its outages. Sounds outrageous? Need not. Because that is what analysts feel about this hyped technology and its equally hyped episodes of outages of the recent past.

Advertisment

Cloud and its outages have been under the scanner for quite sometime. Since 2007, a total of 568 hours of downtime at 13 major cloud services providers had an economic impact of $71.7 million. Not just that, the average down time has been 7.5 hours per year, finds the International Working Group on Cloud Computing Resiliency (IWGCR).

The latest in the outage list are the two recent episodes of Amazon Elastic Cloud Compute’s outage last month and this month and not to forget the ones that knocked off Salesforce as well Microsoft’s clouds the previous month..

But then, cloud providers are not always the only ones to be blamed for such incidences.

Advertisment

“When businesses invest in cloud for most of their requirements, they also need to have some kind of back-up plan for a scenario like this,” says Hansa Krishnamurthy Iyengar, senior analyst of market intelligence at Ovum.

“Outages such as this, though avoidable, are part and parcel of an enterprise’s decision to move to the cloud. Blaming only the provider is not the answer to the situation. Though the provider needs to have a mechanism in place to ensure continuous availability, ideally enterprises need to be prepared to either face an outage or have a solid back-up plan in place to ensure that the end-user continues to get uninterrupted service.”

According to Sanchit Vir Gogoi, associate research manager, Forrester, such incidences are not new, but were happening even in a traditional IT environment.

Advertisment

“It is only that there is a lot of negative publicity around such cloud outages today. It is similar to how you walk into a telecom operator’s service centre and are told that servers are down and so requests cannot be processed,” he explains.

Having said that it does not mean that we should just sit and watch and do nothing about them. An earthquake or flood or lightnings are beyond the control of an ordinary CIO, but then there are certain things which he/she can do to minimize the loss.

Proactive CIOs and robust SLAs

One such thing is auditing, which is not being done due to varied reasons, says Sanchit.

Advertisment

“There are three types of audits that a CIO should be doing before signing up an SLA contract,” says Sanchit. They are:

“People - CIO himself, or a person in his team, or a third party audit body should audit physical levels of security in a cloud service provider's facility.

Processes: Audit on how often is back-up done by such provider.

Technology: What set of hardware, software, middleware, storage, back-up and recovery software do the cloud provider use.”

The once very secretive cloud providers, who used to keep the details of their data centres under wraps,  are today very open about the locations and also permit companies to come and check out such facilities. Today, cloud providers are also confident about sharing information about outages and even provide hour-to-hour details about the measures being undertaken on their service/dash/health board for the users to see.

Advertisment

The cloud industry is evolving and it is time that users put in a little more faith in such services.

Coming to SLAs, 'Cloud/SaaS SLAs need to specifically have a clause that addresses the level to which the cloud provider should be held responsible for revenue loss in case of a failure and most importantly, enterprises need to consider back-up scenarios for when core business systems go down,” adds Hansa.

A factor where CIOs need to take special care is in understanding the SLA document. Both the analysts feel that an SLA expert, who need not be the CIO him/herself, should be a part of the team because SLAs can be tricky at times.

Advertisment

“A standard SLA document needs to be tailored as per the requirement of an organization. It is up to the CIO to actually demand a lot of these things, which a vendor can provide and which might not be a part of the standard SLA document,” notes Sanchit.

And in order to do this CIOs should understand their current and future requirements.

“A lot of time CIOs do not foresee requirement issues and end up signing documents blindly without considering people, applications or workloads or prioritising them on the basis of usage. Mission critical applications need a different and higher set of SLAs in terms of security, availability and back-up, compared to a payroll solution, which can sustain outages,” notes Sanchit.

Advertisment

Now, it is again up to the CIO to specify what he/she wants in terms of back-up, disaster recovery instances, availability etc, which of course means more money.

“Ideally, cloud providers should consider having at least a dual-server failover/load-balanced service as the minimum offering. Today's technological advancements do allow for an automated failover to another data centre, whether domestic or foreign,” avers Hansa.

And, if they do not, demand for it. After all, it is you who is paying and it is your service that is at stake.

“Cloud is not always going to be the cheaper way of doing things, but it gives ease of use in terms of infrastructure managing ability and provisioning,” Sanchit signs off.

smac