The Google Outage: On Misplaced Rage & Crowd Mentality

There is no shortage of outcries over the now infamous 100-minute outage.

“Little ‘dashboards’ aren't going to cut it,” or so it seems [1].

People want answers – and they want accountability.

Just don’t hold your breath for either.

Accountability will be supplied by the market; yes, once you abandon on-premise infrastructure, "it may be cost-prohibitive to return”; but that’s why the onus is on you, during the selection process, to make sure that the move protects against technologies that promote ‘vendor lock-in’ – and there’s plenty of support to manage this risk.

Pardon the terrible expression, but at the end of the day, expertise will converge into the hands of a few who will offer services that the fragmented provides of today can’t – this doesn’t protect against outages (nothing does!) but it does give you the best industry has to offer and delivers it for less than you pay today.

And of course the opportunity to parade around the standard cloud adoption scarecrows wasn’t lost either [2]:

I had a similar reaction to the newly announced Virtual Private Cloud offering from Amazon. Yes, the company has added some security features enterprises desire. But scroll down to the bottom of the FAQ, and in answer to the question "Does Amazon VPC have a Service Level Agreement (SLA)?" you'll see a two-word answer: "Not currently."

And when you think about it -- how could they? The liability would be ridiculously large if Amazon or Google started offering guarantees that came near the value of some unforeseeable disaster. Until that elephant in the living room is dealt with, few enterprise IT folks will take cloud services seriously.

I’d like to know who this mysterious entity that “offers guarantees… near the value of some unforeseen disaster” is? Does you in-house IT offer this? Or does your managed IT provider offer this?

Because this is what I see when I look at managed SLAs, and it doesn’t strike me as unusual:

Service Level Agreement (SLA) Please Note: The maximum service credit(s) available under this SLA in any calendar month shall not exceed fifty percent (50%) of the Customer’s monthly fees otherwise payable in such calendar month. 1. 100% Infrastructure Availability Acme will use its reasonable best efforts to maintain critical infrastructure availability 100% of the time in any given month. If the Customer experiences downtime due to the unavailability of critical infrastructure, Acme will refund the customer 10% of the monthly fee otherwise payable for each 60 minutes of downtime (up to a maximum of 50% of Customer's monthly fee for the affected Service). Critical infrastructure includes functioning of all power and HVAC including UPS. It does not, however, include the power supplies on the Customer's servers unless they are dual power supply machines. Infrastructure downtime exists when a particular server is shut down due to power or heat problems and is measured from the time the monitoring system detects the outage until the time the problem is resolved, the server is powered back on and the monitoring system records the system as active on the network (i.e. pingable). Credits shall not be provided to the Customer if infrastructure downtime is the result of: a) Temporary Service Suspension as per Section 4 of the MSA; b) Emergency Suspension of Services as per Section 6 or the MSA; c) circumstances beyond Acme’s reasonable control, including but not limited to Customer-provided software or equipment or Force Majeure Events; or d) a breach by the Customer of this Agreement or Acme’s AUP.

2. 100% Network Availability Acme will use its reasonable best efforts to ensure that its core network will be available 100% of the time in any given month. If the core network becomes unavailable and the Customer experiences downtime, Acme will refund 10% of the Customer's monthly fee otherwise payable for each 60 minutes of downtime (up to a maximum of 50% of Customer's monthly fee for the affected server). Network availability includes functioning of all network infrastructure including routers, core switches. It does not include distribution layer switching unless the Customer’s service includes redundant connections to distribution layer switching. Also, it does not include software running on the Customer's server. Network downtime exists when all core network interfaces are unavailable as determined and measured by a third party global monitoring system. These reports will be made available to Customer upon request. Credits shall not be provided to the Customer if network downtime is the result of: a) Temporary Service Suspension as per Section 4 of the MSA; b) Emergency Suspension of Services as per Section 6 or the MSA; c) circumstances beyond Acme’s reasonable control, including but not limited to Customer-provided software or equipment or Force Majeure Events; or d) a breach by the Customer of this Agreement or Acme’s AUP.

An interesting exercise is to think about is how quickly your competitors could force your hand on adoption that you may not be ready for: Competing with the Singleminded.

[1] – Gmail Outage Explanation Doesn't Wash http://www.pcworld.com/businesscenter/article/171317/gmail_outage_explanation_doesnt_wash.html?tk=rel_news

[2] – Gmail follies and Google's enterprise pitch http://www.infoworld.com/t/cloud-computing/gmail-follies-and-googles-enterprise-pitch-540