When Gmail went down for a couple of hours this week, I was annoyed but not distressed. As a small business owner, I rely on Google’s paid services for email and calendaring as well as on many of the company’s free services (e.g., Google Analytics). Over the past three years, I have found Gmail to be as reliable and much less costly than our previous in-house, internally-hosted, and professionally maintained (by a full-time IT person) email system.
In monitoring my own calm reaction, I realized that I no longer rely on any single service to keep me online and in touch with the world. Like most of you, I have several email accounts on a variety of email services. I have messaging on my cellphone, Blackberry email and SMS, iTouch WiFi access, Internet phone service (free and paid), and both land line and cable Internet in my home/office and most of the places that I work from while travelling. I also use Facebook, Twitter, and other services to keep in touch with clients, family, and friends. Switching from one mode to the other isn’t really a problem. Not having the history or context available on my laptop or handheld IS a problem, so I’m quite conscious of making sure that I have access to my calendar and contacts (or someone to call or email who can access that info) in a pinch. I only begin to freak out when I am away from any source of Internet/Wireless connectivity. When I visit Uganda to work at the African Rural University, where connections are sparse and slow, I worry about missing something important. (But then I relax and realize that what I’m doing –working with smart young women who are changing people’s lives is MORE important than what I might be missing).
What I noticed in the chatter and punditry that surrounded the news of the recent 2-hour Gmail outage were three main threads of discussion:
- What do you expect? It’s a cloud computing service. Don’t trust anything in the cloud!
- What’s wrong with you guys? Don’t you use alternative emails (Facebook, AOL, Yahoo!) and alternative modes of communication (SMS, iPhone, Twitter)?
- Big deal. It’s no worse than other email services and utility outages on services we have used.
I believe that most organizations—starting with the smallest and moving up to the largest—will soon rely on outsourced/utility/non-hosted computing services and applications. Most of us already do. If you use Salesforce.com, any of the popular email systems, Google for searching the Web, Amazon.com for buying books, etc. you’re USING cloud computing. I trust and rely on these providers to ensure back-up, fail-over, redundancy, and extra capacity to handle spikes and outages. But stuff happens: natural disasters (hurricanes, ice storms), hardware failures, and human error. IT professionals know that you spend the money you need to mitigate the risks you can’t tolerate. There are lots of clouds out there. Spread your risk. Ensure that your cloud service providers make it easy for you to straddle clouds or switch from cloud to cloud if you need to. If you’re running part of your operations in a cloud, you want to control your own security across clouds, which is why I like services like CohesiveFT’s VPN-Cubed.
Distributed, Networked Computing with Lots of Alternatives
Infrastructure Redundancy. One of the things about the Internet that has always seemed almost mystical to me is that the infrastructure that we all rely on is infrastructure that we EACH invest in and maintain. The more Internet servers there are and the more connectivity we buy, the more people can access the Internet faster, cheaper, better. Back to Africa: The best thing that has happened to the Internet in Africa is the advent of the World Cup in South Africa. A new cable has been laid along the east coast of Africa, providing much-needed bandwidth to all of East Africa. A low-orbiting satellite system is being deployed around the equator. Already better and cheaper Internet access is being made available in Uganda, Kenya, and many other countries. These investments are commercially driven by companies that want to reap benefits by delivering entertainment and communication to millions of sports fans around the world (in the case of the cable) and to empower more people in developing countries (in the case of the equatorial satellites). It’s largely a self-healing infrastructure, with bit packets using detours to find their way around outages and dead ends.
Applications Redundancy. Due to the rise of Web 2.0, iPhone and other mobile apps, social media, and SaaS, most people now use dozens of applications in the course of a day. (Remember when we used to use four: word processing, spreadsheets, databases, and email?). Many of us use multi-tabbed browsers with tabs bookmarked to frequently-used services. Lots of us use Blackberries and/or iPhones with multiple apps we can access at our fingertips. This came home to me when I was having trouble reaching a couple of clients by conventional email and phone, so I tried their Facebook emails, and, sure enough, I got almost instant responses. What surprised me is that these are busy senior execs at large corporations. Who would have surmised that they’d be more responsive to Facebook email than their corporate email? But we all have lives, and kids, and family, and hobbies, and passions. So social media has become an increasingly important back channel for reaching people with whom you want to connect.
Service Level Expectations
One of the things I love about my semi-annual “Patty’s Pioneers’” meetings is the opportunity to hang out with a group of people who have been “loosely-coupled” most of their careers. I mean that in both senses of the term.
From an IT perspective, these senior technologists and architects favor loosely-coupled, event-driven, services-oriented, messaging-based architectures. They assume that, at any point in time, something will be down. They design for an environment that is messy, interrupt-driven, chaotic, and unpredictable. They do so by assuming that connections will be broken, processes will be suspended, transactions will be half-completed and then resume. So, through their eyes, a good systems architecture is one in which “stuff happens,” bits are linked to atoms. The virtual world is an abstract representation of the physical world, but physics prevails. However, these folks don’t build large bunkers with layers of walls around them to safeguard their computing strongholds. They build flexible, adaptive systems that dance and improvise. They expect things to break, so they make sure that people can move on and systems can catch up.
The pioneers are “loosely-coupled” in another sense: they reconnect with one another (and with other like-minded souls) over the years and seem to always be able to pick up and catch up with each others’ mental models. (By the way, we’re looking for new recruits to join our band of veteran pioneers, so if you are one of these folks, and you’re not part of our group, drop me an email, tweet, or Facebook ping.)
Cloud computing isn’t unsafe or error-prone. It’s loosely-coupled. So design your life to adapt when one utility or application is temporarily not available. Just switch to another!