Hurricane Sandy Teaches a Lesson About “True” Cloud Computing
This article was written with Mark Famiglietti, the Director of Technology Alliances at Patron Technology. —Gene Carr
Judging by the number of e-mails I got from arts organizations two weeks ago saying their ticketing or website was down, technology infrastructure and disaster recovery have now become two highly critical aspects of running any arts organization. Hurricane Sandy highlighted just how important it is for managers to truly understand technology infrastructure and to make good decisions about it, which is why this article will be more technical than my usual posts.
Simply put, the storm demonstrated that too many organizations are hosting their vital client data, ticket sales, and e-mail on unreliable systems.
How did this happen? As recently as a decade ago, locally hosted systems were the norm. Software was installed on servers in the basement, and it was managed by technically minded people in-house (or on contract), or by full-time database administrators.
But around that time, another technology approach began to emerge, just when we started our company in 2001. We (along with many others) structured our operations very differently. Rather than producing software to be loaded on computers and servers on-site, we began as an “application service provider” — what would later be known as a cloud computing company. Our PatronMail system was hosted, which meant that our clients did not need to install or manage any software. Users logged in over the Internet with any computer browser, and that was it. This remote access is the essence of cloud computing, a trend that has increased dramatically in the past several years.
My colleague Mark Famiglietti sums up what he believes to be the state of many companies’ cloud technology at the moment:
Many technology companies today claim their applications are “in the cloud,” but when you look beyond their marketing talk, they are nothing more than single-tenant applications that lack the important and required building blocks of successful cloud technology: multitenancy, global redundancy, virtualization, elasticity, and scalability. The number of reported outages and business interruptions within the industry as a result of the consequences of Hurricane Sandy suggests that there are many “false clouds” or “thin clouds” in the tech world today.
Here at Patron Technology, we chose to build our PatronManager CRM ticketing/fundraising system on the salesforce.com cloud (platform), an infrastructure that has been developed and fine-tuned over the past 10 years. Thanks to this true cloud platform, we were operational 24/7 during the storm and its aftermath.
The salesforce.com infrastructure runs on three geographically dispersed, mirrored data centers with built-in replication, disaster recovery, a redundant network backbone, and no single points of failure, and processes over 900 million transactions each day for over 3 million users. Today nearly 100,000 Fortune 500 and global businesses trust their business operations to run on salesforce.com, including our 300-plus non-profit ticket-selling clients that are also using this corporate-grade technology.
Mark draws a distinction that I will illustrate for the balance of this article: false clouds versus the true cloud. To do this, I’ve borrowed liberally and edited from the entry on cloud computing on Wikipedia.
Cloud computing has the following two essential characteristics:
- Device and location independence: Users can access cloud-based systems using a web browser, regardless of their location or the device they are using (PC, mobile phone, etc).
- Lower cost: Hardware and other technology infrastructure (such as system backup and data security) are included, and organizations need not purchase, maintain, or manage any servers. Pricing is often based on a “utility” model, where users pay only for the services they use, just as they do with their gas, electricity, or cell phones. And because there is no hardware to manage or software to install, fewer IT staff and skills are required.
But many believe that’s all cloud computing is. And it is in this misunderstanding that the problems lie. “True” cloud systems also employ the following elements, which are essential to delivering the kind of reliability that was not present during the hurricane. I have tried to keep these descriptions as clear as possible, though admittedly some of this is pretty geeky:
Think of an apartment building where all the individual apartments share the same resources for heating, water, and electricity, and yet each individual apartment occupant resides in his or her own unit. As a tenant, you’re not responsible for building your own furnace to get heat — you share the furnace and wires with everyone else in the building, and pay only for what you need. Conversely, if it’s 104 degrees one summer day and you have the air conditioning on full-blast, using drastically more power than usual, you don’t need to do anything to the building’s infrastructure to accomplish that — you just use a bigger slice of the shared resource.
Now, take that same idea into the technology world, and the same idea of centralizing and sharing resources across a large pool of users makes a huge difference.
Imagine that your ticketing system runs on a “true cloud” platform in a multitenant environment. You usually sell about 100 tickets each day, but there’s a special event coming up: it’s 10 AM on Wednesday morning and your season is about to go on sale. There’s going to be a huge number of people trying to buy those tickets, which means a lot more traffic than usual to the system. In a multitenant environment, servers and data storage devices (i.e., the machines that run the ticketing system) are shared by many clients, which means that the infrastructure can handle this dramatic increase in demand. Your organization doesn’t need to pay for a bunch of extra servers to handle the peak load capacity of this one very high-demand day; in the cloud, you can simply increase or “burst” capacity as needed.
Cloud companies monitor performance and adjust resources efficiently to provide “elasticity.” Simply put, the manager (or the system itself) can grow the infrastructure by adding more machines, or take machines offline when they’re not needed. If your production schedule changes and you’re going to be presenting high-demand concerts all the time, and thus need that higher capacity regularly, the system can grow accordingly without your having to buy new equipment.
True cloud systems have multiple and redundant data centers in operation. This is a critical aspect of cloud computing, and here’s where the hurricane demonstrated the flaw in many systems. In some cases, all the servers were located in one location, with no geographic distribution. Even if the system offered real-time replication of data to multiple machines, having those backup machines in the same location as the main one doesn’t help if the building gets flooded. A true cloud environment has many data centers in different locations, which is why corporate America is choosing to move to the cloud — these systems are especially suitable for business continuity and disaster recovery.
According to Zeus Kerravala, founder and principal analyst with ZK Research (a networking and cloud research firm), “One of the most understated use cases for the cloud is disaster recovery. The cloud is built for backup and recovery, with geographically disbursed networks.” And Michael Cohen, SVP of Cloud Sherpas, a Google partner company that services Google Apps and other products, says, “Google has had few problems of network availability. With Google, data doesn’t live in any [one] data center,” Cohn said. “It’s fragmented and distributed in multiple data centers.”
It’s clear we’re moving into a technology world dominated by cloud computing, and one of the effects of this is that lots of companies will claim to operate in the cloud. It makes good business sense, when you’re looking at new technology for your organization, to ask the suppliers to describe exactly how their operations stack up against each of the items I’ve laid out above. And, importantly, I hope you put reliability and disaster recovery at the top of your list. The technology platform you choose probably is the most important decision you will make, and everything else is secondary.
At this point, I am guessing you may feel that you don’t have the technology background (or the interest) to assess systems in this manner. But that doesn’t mean you shouldn’t do it. Just as you hire lawyers and accountants to help you run your business with their unique expertise, you should involve knowledgeable board members, friends, or even consultants to help advise you in this arena.
As we recover from Hurricane Sandy, I hope we can all learn technology lessons that will put our industry in a better place to weather future crises. Organizations like yours find it difficult enough to operate as they are, without having the concern that come the next storm they could be out of business for days or weeks.
– – – – – – – – – – – – – – – – –
Speaking of the cloud, check out these free web tools!
– – – – – – – – – – – – – – – – –