Major service disruption hits Salesforce.com

Major service disruption hits Salesforce.com

Gremlins in the works at Salesforce.com as a Fourth of January outage leaves subscribers unable to use the platform for over an hour.

Software as a Service (SaaS) provider, Salesforce.com experienced a near-global service disruption for over an hour on 4 January. The downtime came as the company dealt with the highest number of transactions for weeks, as subscribers returned from their Christmas breaks.
The disruption affected all of Salesforce.com’s instances, including Japan, North America, Europe, Middle East, and Africa. Luckily for the company the downtime hit between 12:10pm PST to 1:25pm PST (8:10pm GMT to 9:25pm GMT), after office hours for many global regions.
A second, shorter disruption was recorded within a half hour of the previous outage, but lasted just seven minutes.
Salesforce.com is yet to issue a statement on the disruption, but according to Trust status website, the service “experienced a double failure which resulted in access to the storage subsystems to become unavailable. The Salesforce.com team had to reboot systems to restore connectivity."
December and January have in the past proved challenging months for the SaaS company, which has experienced several outages during the winter period in recent years. Major outages affected the service in December 2005 and January 2009, with separate minor outages recorded in January 2006.
A major disruption to the service also occurred in April 2006, while several Cloud platforms, including Salesforce.com and Amazon’s EC2 and S3 environments, together with major ecommerce sites like Walmart, were also affected by a Denial of Service (DDoS) attach on 23 December 2009. The attack affected users of NeuStar’s ultraDNS service, which translates web addresses into network-friendly Internet Protocol (IP) addresses.
The latest outage is likely to once again raise the debate over the pros and cons of Cloud services, and the question of whether or not the Salesforce.com disruption will derail the company - or indeed the SaaS movement - has already been posed by some quarters.
As has been well-documented in the past, such incidents have demonstrated the possible vulnerabilities of relying on a single vendor for IT support and services. However, proponents of the Cloud model highlight that most organisations are equally as susceptible to operational problems - if not more susceptible - than the Cloud vendors.

Comments

As a competitor to Salesforce, and a SaaS provider you would guess that I would come down on the side of the SaaS model, but actually, wouldn’t anyone.... even though Salesforce are a competitor their track record given the numbers of users they support is pretty impressive.

It is a little like getting a flat tyre and then suggesting that everyone debunk back to horse and cart.

I am not underestimating what an outage like this can do to your customers businesses, but when taken in context it is a small amount of downtime. I wonder how long 'on premise' solutions where out of action in the last 3 years (using the timeframe that you quote in your article), I would guess it could have been a great deal more than a couple of hours.

As I say, for balance, we are a hosted solution provider (http://opencrm.co.uk) ourselves, but babies and bathwater do spring to mind.

Graham

Graham Anderson

Software Add-ons - The Home of OpenCRM

SF are not unique in suffering outages - google (G-Mail), Microsoft (Azure + Hotmail etc..), Blackberry, Rackspace and Amazon have all suffered outages at some stage. However which on-premise systems have been 100% reliable over a period of years? Not any I'd guess.

Nick Carr in the Big Switch is right in that people who deny the coming of cloud computing to the masses are like the Victorians who were not convinced of utility grid supplied electricity - all eras come to an end. The end of departmental and corporate data centres is nearing it's end and the winners in Cloud computing will be those which can provide more reliable, cost efficient and scalable services.

TriSys has been offering cloud computing to the recruitment industry for over 8 years now and during that time our uptime has exceeded that of traditional on-premise IT systems. We would certainly like 100% availability to our customers, but they are generally happy with four-nines.

Garry Lowther, CEO, TriSys Business Software, www.trisys.co.uk

 

 

Brian Keedwell, Integrated Marketing

If the architecture is high integrity distributed databases, with bi-directional asynchronous replication (driven by such as iAnywhere), the consequence of outages ANYWHERE IN THE SYSTEM is negligable.

Let's put this in context. Although many SaaS vendors strive for 99.999% availability, few achieve it, the best are only 99.99%. 99.999% equates to 5 minutes down per year, 99.99% one hour. So for an SaaS vendor Salesforce.com at 99.99% (same as us!) are up there with the best.Then compare this with most internal IT systems' uptime. Few get away without at least three or four day's down time a year - that's 99%, and many consider being down outside of business hours (which this was, for UK users) as completely acceptable.

It's just that people want SaaS systems to be perfect, 100% available. Which is a great goal but unrealistic today. A one hour glitch is unfortunate, but it doesn't destroy the validity of SaaS's main claim of being one heck of a lot better than in-house systems.

By the way, I notice that the Barclays banking web site is often down for an hour during business hours, maybe as much as once a month. Haven't seen anybody wanting to leave online banking so far.

John Patersonwww.reallysimplesystems.com

 

Back to top Back to top