This post was originally published on this siteI posted a few months back about ESXi queue depth limits and how it affects performance. Just recently, Pure Storage announced ...
Founded in 1872, The Boston Globe is one of the most respected and decorated daily newspapers in the United States. Hauling in 23 Pulitzer Prizes since 1990, the publication is well-known for its Spotlight Team that exemplifies the power of investigative reporting.
A great deal has changed in the publishing world since The Boston Globe’s founding 144 years ago. I recently sat down with Wade Sendall, Vice President of IT of The Boston Globe, to discuss what Hello Possible means for his business.
JM: As a news organization, how has your business changed since the turn of the century?
WS: Let me start by saying what hasn’t changed. As a news organization relied upon by our readership for timely and accurate information that affects their lives, we continue to live in a deadline-driven business. That said, the days of running an analog, print-only business are well behind us. This is not to say that we don’t maintain a significant revenue stream from our print editions; but we are decidedly moving in the direction of a digital subscription model supported by highly dynamic web content.
JM: How is this changing your business operations?
WS: In a word, “dramatically.” In effect, we’re now an organization comprised of many different companies operating under one umbrella. Given the 24×7 news cycle we now live in, we maintain B2B, B2C, logistics, manufacturing, product development and web development operations.
JM: Given these significant changes in the structure of your business operations, what are the key initiatives of The Boston Globe?
WS: The tip of the spear is “increased readership”. With the wholesale shift from print to digital content, we are concentrating our resources and efforts on expanding our digital subscription model significantly. And we’re doing a good job of that in view of the fact that we enjoy one of the largest digital subscription models in the country. This is important for a couple of reasons. The revenue share model between print and digital has changed completely. We used to see a 4:1 revenue stream from advertising vs. circulation on the print side. That has been flipped on its head. We are fundamentally changing this model by going from ad-based to circulation-based income.
JM: How is this market dynamic impacting your IT operations?
WS: Like any IT organization servicing a large enterprise, we are tasked by management to implement technologies and business models that reduce CAPEX and OPEX overhead. At the same time, we’re asked to optimize support maintenance costs – without sacrificing our ability to achieve our core business objectives. Because we run 24×7, maintenance windows have shrunk considerably. This mindset was a major influence in our evaluation of a new storage infrastructure that was necessitated by a normal refresh cycle – due to an aging system – and continuous growth in the volume of data we support.
JM: As a Pure Storage customer, can you explain the selection process for all-flash arrays?
WS: Let me start by saying that from a performance standpoint, making the transition from spinning disk to flash storage is a no-brainer. The one proviso? The cost basis had to be competitive, and my team is pretty cynical and skeptical of vendor promises. That’s why the Pure Evergreen program – that provides upgrades for the entire life of the one-time purchase price of an array – blew us away. There’s nothing else like this in the industry. What we love about this model is that we know the cost of an array until the end of life. Let me put it another way: the finance team loves it when IT expenditures are flat and predictable.
JM: Aside from cost considerations, what other criteria was involved in deploying Pure?
WS: Let’s start with performance. Every application core to running our business is accessed by XEN-based VMs running on Pure arrays. That covers everything from newsroom apps to content management, digital production, web-authoring and VOIP. So we needed robust I/O performance. After years of running our operations on spinning drives and NAS storage, the difference in I/O performance was simply night and day once we cut over to Pure. One of the most notable examples is the performance of our Shoretel VOIP system that now runs 8 – 10x faster on Pure. And that’s proven to be a huge advantage in handling customer support calls.
JM: You pointed out that your system administration team is skeptical of vendor claims. How did Pure win them over?
WS: Every vendor tries to sell the concept of simplicity. So after I attended a conference where one of the founders of Pure told a story about bringing an array home to his 8-year old kid to prove how easy it was to cable the device and make it operational, my guys had serious doubts. Those doubts were quickly put to rest. I don’t mean to sound pejorative when I say this, but Pure has dumbed things down when it comes to installing an array – right down to plugging in colored-coded cords. Turns out the kid was right. We were up and running inside of three hours during our proof-of concept testing.
JM: Any other advantages that won over your IT skeptics?
WS: Our team all agrees that writing to flash memory is much less volatile than mechanical storage devices. But because of our extensive use of VMs, our sys admin team was skeptical that the Pure arrays could deliver deduplication levels to lessen the capacity impact of our workloads. We have roughly 12 TB of storage, but our VMs “think” they’re handling 43 TBs. Once we migrated our workloads to Pure, the compression ratios were off the chart. While our IT team assumes there are some skeletons in the closet with any product, they haven’t found any yet with the Pure arrays. The cynics are becoming true believers.