Cloud Latency, Enterprise Storage and Speed-Of-Light Limitations
With data-storage needs expanding rapidly, companies have been forced to repeatedly expand the footprints of their primary, backup and disaster-recovery storage infrastructures. Why shouldn’t they instead work with MSPs to tap into the elastic capacity of the cloud? It seems like a no-brainer. And yet, most enterprises either don’t put their primary data in the cloud or struggle once they do.
While many people attribute reluctance about cloud adoption to concerns about security or control, less discussed is an equally serious hurdle: latency. Data travels at the speed of light, but users and software still feel a lag when the data has to move across significant distances and complex networks between employees and cloud infrastructure. This is especially problematic when enterprises use the public cloud to store data while also maintaining compute workloads on premises.
When I talk to industry analysts about this issue, they tell us that as many as half of their customers that have gone to the cloud bring their workloads back in-house because of unacceptable latency and performance. How does that reflect on a channel partner who advocated the benefits of cloud?
Blame Physics
In his book on the cutthroat financial services industry, “Flash Boys,” Michael Lewis writes about one Wall Street firm that invested in an 827-mile cable between Chicago and New Jersey that cut through mountains and traveled under rivers. The extreme (and extremely expensive) project was designed to cut latency by 4 milliseconds — enough to make a competitive difference in the trading industry. For the vast majority, that kind of investment is unfathomable. Yet Amazon, Google, Microsoft and other cloud providers build their massive data centers in regions where they can scale at a reasonable cost and have access to large power grids. Usually, that puts the public cloud hundreds or thousands of miles away from customer locations. This is less of a problem when companies want to store massive volumes of archival data or use the cloud for backup or disaster recovery, since users aren’t expecting instant, frequent access to that data. Business applications that leverage primary data and require less than 10 milliseconds to work properly, however, are a different story. Users expect real-time access to hot data, making these workloads extremely performance-sensitive. When data is further away, latency lags get worse.
Making the cloud work for data storage management, then, means partners need cost-effective solutions for latency before they recommend the cloud for storage management. In pursuit of that goal, some channel partners work with customers to beef up their network infrastructures, rebuild applications or seek out more efficient traffic routes enabled by cloud exchanges or network hubs in major colocation sites. While these options are more feasible than digging an 827-mile route through the mountains, they are still capital-intensive and don’t free IT teams or their advisers from managing complex infrastructure.
Before enterprises take full advantage of the public cloud model – at least in terms of what it can do for workloads with stringent performance requirements – they need to see scenarios in which cloud storage looks, walks and talks like local storage. That means partners need to start looking for solutions that approach storage management in entirely new ways.
Seek service providers that can leverage direct connections and WAN optimization techniques on the front end while helping to manage infrastructure on the back end. Service-based monitoring tools, which can help improve performance of Web apps, cloud apps and every level of your IT stack, provide a snapshot of your customer’s full environment and further improve infrastructure management. However, enterprises will see the greatest returns from working with service providers who can help them step away from infrastructure-demanding processes — a trend we expect will gain traction in the months to come.
Ellen Rubin is an experienced entrepreneur with a track record in leading strategy, market positioning and go-to-market efforts for fast-growing companies. She is the chief executive officer and co-founder of ClearSky Data, an early-stage company that recently launched out of stealth mode. ClearSky Data’s global storage network simplifies the entire data lifecycle and delivers enterprise storage as a fully managed service. Most recently, Rubin was co-founder of CloudSwitch, a cloud enablement software company that was acquired by Verizon in 2011.
