Skip to main content

In Conversation with Newsweek's CIO: "The Cloud Was a Good Move for Us."

By June 8, 2010Article

It is not an exaggeration to say that Joseph Galarneau, CIO of Newsweek, is on the leading edge of the cloud revolution: Under his leadership,’s mission-critical infrastructure went live on Amazon Web Services two weeks ago.
Joe is the Senior Vice President of Operations and Chief Information Officer at Newsweek. When he wears his CIO hat, Joe is responsible for all technology at Newsweek including online infrastructure, online application development, and project management associated with the web site. In addition, he is responsible for technologies that run Newsweek’s internal business and publishing systems. In his SVP of operations role, he also oversees manufacturing, distribution, digital ad operations, print ad operations and real estate.
I spoke with Joe about his experience working with a public cloud infrastructure like Amazon, and the impact of the digital revolution and cloud computing on the media and publishing industry.
Here are some excerpts from the discussion:
How far along is the publishing industry on its transition to digital media?

What we are seeing now is that a lot of consumer media companies are generating significant revenues from digital operations. It’s hard to say when — I wish I knew — but I think we will soon reach an inflection point where we have to ask what’s happening to print and is there erosion of print growth vs. digital.
I used to work for a business-to-business publishing company, which has completely inverted their revenue model going from 95% print to 95% digital over the past ten years. However, that transformation resulted in the company plummeting to less than half of its original size. In general, I think digital publishing will drive down the revenues and size of the industry in next 5-10 years.
I don’t think print will disappear completely, but I wouldn’t hazard a guess as to what percentage of print would be left in 10 years from now, particularly as consumer publishers are just now starting this transformational journey. But looking at the overall downward trends, clearly it will be less than what it is today.
People are struggling with digital monetization models — should we charge (and if so how much) consumers for iPhone apps vs iPad apps, the role of paywalls, etc. There will a shakeout in our industry in the next couple of years regarding what consumers are willing to pay and what our ultimate economic model will look like.

How is the role of technology changing in the new world of digital media and publishing?

Digital is a big thing for most of us and is responsible for an increasing share of our revenues. Over the years, we have been making increasing investments in technology because that is becoming the new distribution platform be it the Web, be it mobile or tablet devices.
If you look at the digital side of Newsweek, it is not unlike a pure-play dot-com in terms of how we view technology as a strategic enabler and differentiator for that side of the business. That said, since we are a relatively small company, we take a realistic view of what we should be doing ourselves and what we should let other companies do for us in terms of technology build-out.

Given the mission-critical role of technology in your digital operations, how did you come to the decision to go “all in” and deploy on a public cloud like Amazon?

Historically, was always hosted on servers that were not owned by Newsweek itself. Our servers were hosted for many years by MSNBC and more recently they were hosted by our sister company, The Washington Post.
We also newly acquired a highly-scalable content management system (CMS) with a more modern architecture that was different from anything else in use at The Post. That gave us an opportunity to look at how we can host this differently. We didn’t own any hardware; we had no datacenter; so in essence we were really taking a green field approach.
Also, we have an ongoing partnership with MSN and MSNBC where the sites promote our content several times a week on their homepage. When that happens, we can see traffic levels go up by as much as 20-30 times (we have also seen bursts of two orders of magnitude) in the course of a few minutes. This partnership makes our traffic highly “burstable.” It would be extremely expensive if we had to build our own infrastructure to support the peak traffic loads.
These factors lead us to seriously consider the cloud for our backend infrastructure. We did internal studies and risk analysis and determined that the public cloud was the way to go — particularly when we discovered that we could easily save 20-30 percent on the infrastructure costs (and this is not even factoring in the burstable aspects of the traffic.) Since we were starting from scratch, the question before us was why invest in any assets internally.
Once we made the decision to move to the cloud back in fall last year, we pursued that strategy aggressively. We did a series of pilots which were very successful and then continued on with a few other initiatives (photo galleries and so on) and then finally did the entire site re-launch on Amazon Web Services (AWS) in late May. Prior to the site re-launch, we had served more than 100 million page views and had 14 million visits over the past seven months with our AWS experiments.

What has been the cloud’s impact on staffing and costs?

We realized very early that when you are recruiting technical talent for doing system administration in the cloud, you need to look for people that are more like solution architects and less like traditional system administrators. You need people with development skills that can deal with architecture and with this big bucket of tools and the abstractions that the cloud offers. So the staffing of our SA team has been challenging.
If I were to build my own data center, I would need to worry about the entire “gut-level” infrastructure, such as building my own private network with routers, switches, firewalls, load balancers, cables, racks, power, cooling, and so on. Amazon pricing includes almost all of the management of this network architecture.
When we get to the level of servers, and consider the costs of building our own fault-tolerant resilient architecture with built-in scaling, multiple geographical zones and disaster recovery, that’s where you get into hard cost savings.
We are working to fine-tune our capacity and demand needs on Amazon using a proper blend of reserved instances, on-demand instances, and spot instances. This requires good engineering to get you to really cost-efficient operations. We are squeezing labor costs by more automation, smarter scaling, and more economical usage of storage. It’s not a slam-dunk economically but we can easily see TCO savings of 20-30 percent and I think if you are starting from zero, [the cloud is] a super compelling proposition.

What are some lessons you learned during your experience moving to Amazon?

You have to approach application engineering in the cloud differently. Essentially, you have a nearly unlimited palette to design your infrastructure so it can get overwhelming for infrastructure people who are not used to working at this higher level of abstraction. Because the boundaries are blurred, you also need to be clear about who calls the shots regarding web operations and infrastructure architecture (although that’s not a problem here because the same group handles both.)
The Amazon elastic load balancing functionality isn’t as mature as what you would get from traditional networking vendors like F5 so we also had to work around that. We also had to use a third-party company to obtain DNS services which aren’t offered on Amazon. CloudFront, their CDN service, has good basic capabilities but needs some work too, so we stayed with our current CDN provider. All of this may change in a year or two because we see that Amazon is very responsive to our needs and is learning quickly how to work with the enterprise to provide a broader and deep set of product offerings.

What about security, data privacy, performance and other risk issues? is a public facing website. Clearly, we don’t want people to hack into our site, but then we are not putting super-sensitive business or personal data out there in the cloud. I think we have built a hardened infrastructure on Amazon that is suitable for this application.
However, the website is mission-critical: we serve over 10 million visitors per month and we can’t be dark for very long. Amazon’s published SLAs are less exacting than that of other non-cloud hosting vendors. You could look at that say, “That is not good enough for me — I need ‘five 9s’ reliability for my mission-critical applications.” We took an in-depth look at how Amazon architected their infrastructure. We discovered that if we engineered our own fault-tolerant and resilient layer on top of Amazon, we can get to whatever availability we need. It’s definitely surmountable by just being smarter about it. Our architecture and Amazon’s infrastructure is holding up nicely without any noticeable hiccups. Every week we are learning and getting smarter about building-in more fault-tolerance and resilience into our architecture. Our goal is to be really bulletproof by the end of the summer.
Regarding performance, we certainly considered that in our risk analysis, but we never saw any data that would lead us to believe that would be an issue in practice. We actually experienced good improvements in our response times. Since we also switched to a new product and application architecture, it hard to say where the improvements came from. That being said, Amazon could partner with ISPs and CDNs to provide better performance to our end-users.

Any final thoughts for other enterprise CIOs considering a move to the cloud?

People talk a lot about lock-in, reliability, and security in the cloud. I think all of these things are surmountable with good engineering and good planning. It’s not really all that scary.
Now clearly, there are shortcomings — everything has shortcomings — but I think if you build a smart team and you go into this with your eyes open, I’m very convinced that the cloud is revolutionary in how we think about application delivery and infrastructure.
We have had an absolutely positive experience with [the cloud] and I’d do it again in a millisecond. Quite frankly, the cloud was a good move because as a CIO, it gets me out of projects that are not “value adds” for our business.

Upcoming Cloud Event: We are co-hosting a Software Executive Networking Dinner “Achieving Technology Leadership in the Cloud” on Tuesday, June 29 at 6:00 pm at Perry’s steakhouse in Austin, Texas. If you are based in the Austin area or are visiting, please join us for the dinner and enhance the quality of the interactions and networking. Please drop me a note if you can make it to the dinner.
Kamesh Pemmaraju heads cloud research at Sand Hill Group. He welcomes your comments, opinions, and questions. Drop in a line to

Copy link
Powered by Social Snap