New Requirements for Cloud Service Providers: Turning IT Costs into Profits

According to Wikibon research, IT spending as a percent of revenue at Global 2000 companies has dropped from 4-5% in 2000 to 2-3% today. In a long-term trend that started in 2001, traditional IT shops have suffered through massive budget cuts and a mandate of “Do More With Less.” Meanwhile, cloud service providers (CSPs) have been investing in cloud infrastructure, growing like crazy and innovating; becoming a bellwether of IT innovation.

As the chart below shows, traditional IT organizations have lived on a steady diet of budget cuts while CSPs are investing and beginning to flourish. This “innovation gap” will likely continue, as CSPs are profit-driven whereas the vast majority of IT organizations are viewed as a cost center. Profit centers invest; cost centers cut.

Traditional services organizations including CSC, IBM, HP, etc. are looking hard at this trend, as are many traditional IT shops. Their business models are under fire and many have lived off of a break/react/fix mentality. These organizations will increasingly come under margin/budget pressure as more work is done in the cloud and the adoption of automation escalates; reducing the need for armies of humans both in services companies and in IT shops.

Most traditional IT operations are ill-equipped to deal with this trend because of the numerous organizational barriers, including a stove-piped management structure with fiefdoms to protect and many hundreds of applications to manage across the portfolio. For CSPs on the other hand, it’s a bottom line exercise with a laser focus on those activities that will drive profits.

IT is a Profit Center for CSPs

The main premise of Nicholas Carr’s famous 2004 book “Does IT Matter,” was that IT is of diminishing importance and technology investments no longer provide competitive advantage. While this theory seems largely applicable to companies with traditional IT, the hypothesis has been proven incorrect for large Internet firms including Google, Amazon, Facebook and a new breed of emerging cloud service providers. Clearly technology has been the mainspring of innovation at these organizations and has led to the creation of hundreds of billions of dollars in market value.

The bottom line for CSPs is IT and the people and processes around technology, remain a competitive advantage. Specifically, from the standpoint of a CSP:

  • Leading cloud technologies translate into profitable business models. There is a major difference between traditional IT, traditional hosting and elastic compute and storage that is for sale.
  • To traditional IT organizations, ROI means cutting costs to get a return (TCO); rather CSPs are driven by the numerator in the ROI equation more than they are cutting TCO.
  • CSPs are advancing metrics like revenue and profit per customer. These are the KPIs that increasingly drive CSPs beyond metrics that track how many times the light of the server goes red.
  • CSPs are focused on the lifetime value and total margin that a solution can deliver.
  • CSPs are looking for suppliers that can offer a net plus to their bottom line. They are willing to take risks to get big returns. Most IT organizations, by their very nature are risk averse and typically look for suppliers that will help them maintain their jobs.

This is not to say CSPs don’t worry about costs—they do. But they will invest in technologies that will drive profits. A good example is automation—an important component of a CSPs profit engine. As David Floyer recently wrote in a Wikibon Professional Alert, in addition to investing in capital equipment, cloud service providers are going hard after automation. Google and other Internet giants have shown the way in the past decade as building massive scale infrastructure to support hundreds of millions of users required nearly full automation of IT operations. It was simply too expensive and too complicated to manage infrastructure in a brute force manner using humans.

To be specific, the basic premise for these large scale operations is that without automation and complete management of each component of the entire infrastructure – to include servers, storage and networking – cloud systems won’t scale; period.  We’re not just talking about basic systems management tasks such as provisioning. We’re including full maintenance of the system– e.g. microcode updates, problem determination, data recovery, etc. In other words, all the tasks that a Systems Engineer used to perform must be automated in the cloud and managed by machines, not people. The one exception is the replacement of physical hardware…but that’s about it.

What It Means for Storage

The cloud needs better storage; period. What does ‘better’ mean? In this context it means simpler to manage, less complex and more efficient. This applies for archival storage – e.g. object storage like S3 and block-based storage such as Amazon’s Elastic Block Storge (EBS). As I’ve written in the past, better for many customers also means more reliable with SLAs that are enterprise class. This is the entire premise behind companies like Nirvanix, which offers improved SLAs for enterprises for object storage.

Do CSPs and their enterprise customers need a more robust and reliable analog to EBS? In my view, yes. To date, cloud storage has largely been about supporting Web-based applications running on non-relational databases. There is an emerging breed of applications CSPs are readying for the cloud and they are typified by traditional systems running relational database management systems with large data requirements. To date, cloud storage hasn’t been able to accommodate these systems—which is where EBS is targeted.

As with S3, we believe there is demand for enterprise grade block storage and many organizations will pay a premium for better SLAs than Amazon will offer. The question remains—what type of block storage will backend these systems?

Answer – probably not traditional RAID storage systems that are running in the enterprise today. Yes the big storage whales will get their fair share of this business – because they are large and trusted. But generally these systems have an uphill battle in the cloud for four reasons:

  1. They lack the ability to guarantee quality of service (QoS) at the application level because the gap between server and storage performance continues to grow. Onerous spinning disk rebuild times exacerbate this problem. Most traditional RAID systems deliver guaranteed capacity, not performance (see All-flash Arrays).
  2. Generally, traditional storage systems aren’t optimized for mixed workloads typified by the cloud. Most of these platforms were designed 10+ years ago; they’re not optimized for SSD, lack QoS capability, multi-tenancy, or management automation.
  3. These systems typically lack modern data optimization/reduction techniques. While 3PAR’s thin provisioning is a leading example, most of these systems, including 3PAR, shun in-line de-duplication and compression because of performance concerns.
  4. Traditional RAID systems lack cloud-specific APIs that, as Floyer’s research points out is critical to automation.

Here’s the prediction:

All-flash arrays will become the de facto standard for true enterprise cloud-based block storage by 2015. By ‘true cloud’ we mean elastic storage, pay-by the drink and enterprise SLAs.  These all-flash systems will be less expensive than FC and SAS-based spinning disk arrays and ultimately dominate the cloud landscape for block-based storage.

If you’re looking for a trendsetter in this field keep an eye on SolidFire. As we talked about last week on the Cube, the company just did another raise and is exclusively focused on CSPs—it’s their only channel. Here’s a video clip from that discussion:

If SolidFire can execute and deliver a reliable product, observers should expect the company to have a rapid ascendency in the marketplace. It has all the bells and whistles you’d want in a block-based cloud storage device and with a CEO that comes from Rackspace it understands well the CSP marketplace and requirements. Expect that like 3PAR before it, the company will see many copycats. In fairness companies like Pure Storage also have the product attributes but the difference is SolidFire is exclusively positioning for the cloud.

While it can be argued betting solely on CSPs will limit the company’s TAM, given the trends discussed in this post—it’s a bet worth placing.